feat: enhance text protocol conversion and advanced custom routing (#5825)

* refactor: consolidate relay protocol converters * refactor relayconvert text converters * feat: refine relay converters and advanced custom routing * refactor: enhance logging and add thought signature handling for Gemini requests * refactor: enhance channel cache and pricing endpoint handling for advanced custom models * feat: preserve billing usage semantics * feat: add protocol-aware billing usage * Delete useless files * chore: update action versions in workflow files * chore: update Docker action versions in workflow files * fix: harden billing usage settlement and hot-path route matching - estimate Gemini completion tokens locally when billable usageMetadata is prompt-only but output content was received (e.g. client aborts the stream before the final chunk), and rebuild the attached billing_usage as estimated so settlement does not bill zero output tokens - guard NewClaudeMessagesBillingUsage against all-zero ClaudeUsage, matching the OpenAI/Gemini constructors, so a zero billing_usage cannot override a non-zero top-level usage during settlement - cache compiled advanced-custom route model regexes; they run on the request hot path and were recompiled per request - move the effectiveBillingUsage remap to PostTextConsumeQuota only, and document that calculateTextQuotaSummary expects remapped usage - document the updatePricingLock -> channelSyncLock lock ordering that InitChannelCache/CacheUpdateChannel rely on, and the aux-struct pitfall in GeminiChatResponse.UnmarshalJSON

feat: enhance text protocol conversion and advanced custom routing (#5825)
* refactor: consolidate relay protocol converters * refactor relayconvert text converters * feat: refine relay converters and advanced custom routing * refactor: enhance logging and add thought signature handling for Gemini requests * refactor: enhance channel cache and pricing endpoint handling for advanced custom models * feat: preserve billing usage semantics * feat: add protocol-aware billing usage * Delete useless files * chore: update action versions in workflow files * chore: update Docker action versions in workflow files * fix: harden billing usage settlement and hot-path route matching - estimate Gemini completion tokens locally when billable usageMetadata is prompt-only but output content was received (e.g. client aborts the stream before the final chunk), and rebuild the attached billing_usage as estimated so settlement does not bill zero output tokens - guard NewClaudeMessagesBillingUsage against all-zero ClaudeUsage, matching the OpenAI/Gemini constructors, so a zero billing_usage cannot override a non-zero top-level usage during settlement - cache compiled advanced-custom route model regexes; they run on the request hot path and were recompiled per request - move the effectiveBillingUsage remap to PostTextConsumeQuota only, and document that calculateTextQuotaSummary expects remapped usage - document the updatePricingLock -> channelSyncLock lock ordering that InitChannelCache/CacheUpdateChannel rely on, and the aux-struct pitfall in GeminiChatResponse.UnmarshalJSON
c36418c8 · Calcium-Ion · GitHub · 1250fb2e · c36418c8 · c36418c8
Commit c36418c8 authored Jul 11, 2026 by Calcium-Ion Committed by GitHub Jul 11, 2026
106 changed files
--- a/.github/workflows/docker-build.yml
+++ b/.github/workflows/docker-build.yml
@@ -36,7 +36,7 @@ jobs:
    steps:
      - name: Check out
-        uses: actions/checkout@v4
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
        with:
          fetch-depth: ${{ github.event_name == 'workflow_dispatch' && 0 || 1 }}
          ref: ${{ github.event.inputs.tag || github.ref }}
@@ -59,23 +59,23 @@ jobs:
          echo "Building tag: ${TAG} for ${{ matrix.arch }}"
      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
+        uses: docker/setup-buildx-action@bb05f3f5519dd87d3ba754cc423b652a5edd6d2c # v4.2.0
      - name: Log in to Docker Hub
-        uses: docker/login-action@v3
+        uses: docker/login-action@c99871dec2022cc055c062a10cc1a1310835ceb4 # v4.3.0
        with:
          username: ${{ secrets.DOCKERHUB_USERNAME }}
          password: ${{ secrets.DOCKERHUB_TOKEN }}
      - name: Extract metadata (labels)
        id: meta
-        uses: docker/metadata-action@v5
+        uses: docker/metadata-action@dc802804100637a589fabce1cb79ff13a1411302 # v6.2.0
        with:
          images: calciumion/new-api
      - name: Build & push
        id: build
-        uses: docker/build-push-action@v6
+        uses: docker/build-push-action@53b7df96c91f9c12dcc8a07bcb9ccacbed38856a # v7.3.0
        with:
          context: .
          platforms: ${{ matrix.platform }}
@@ -90,7 +90,7 @@ jobs:
          sbom: true
      - name: Install cosign
-        uses: sigstore/cosign-installer@v3
+        uses: sigstore/cosign-installer@6f9f17788090df1f26f669e9d70d6ae9567deba6 # v4.1.2
      - name: Sign image with cosign
        run: cosign sign --yes calciumion/new-api@${{ steps.build.outputs.digest }}
@@ -117,7 +117,7 @@ jobs:
        run: echo "TAG=${{ needs.build_single_arch.outputs.tag }}" >> $GITHUB_ENV
      - name: Log in to Docker Hub
-        uses: docker/login-action@v3
+        uses: docker/login-action@c99871dec2022cc055c062a10cc1a1310835ceb4 # v4.3.0
        with:
          username: ${{ secrets.DOCKERHUB_USERNAME }}
          password: ${{ secrets.DOCKERHUB_TOKEN }}

--- a/.github/workflows/docker-image-branch.yml
+++ b/.github/workflows/docker-image-branch.yml
@@ -21,7 +21,7 @@ jobs:
      contents: read
    steps:
      - name: Check out branch
-        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
        with:
          fetch-depth: 1
          ref: ${{ inputs.branch }}
@@ -68,7 +68,7 @@ jobs:
      id-token: write
    steps:
      - name: Check out branch
-        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
        with:
          fetch-depth: 1
          ref: ${{ needs.prepare.outputs.sha }}
@@ -79,24 +79,24 @@ jobs:
          echo "Publishing version: ${{ needs.prepare.outputs.version }} for ${{ matrix.arch }}"
      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@8d2750c68a42422c14e847fe6c8ac0403b4cbd6f # v3
+        uses: docker/setup-buildx-action@bb05f3f5519dd87d3ba754cc423b652a5edd6d2c # v4.2.0
      - name: Log in to Docker Hub
-        uses: docker/login-action@c94ce9fb468520275223c153574b00df6fe4bcc9 # v3
+        uses: docker/login-action@c99871dec2022cc055c062a10cc1a1310835ceb4 # v4.3.0
        with:
          username: ${{ secrets.DOCKERHUB_USERNAME }}
          password: ${{ secrets.DOCKERHUB_TOKEN }}
      - name: Extract metadata (labels)
        id: meta
-        uses: docker/metadata-action@c299e40c65443455700f0fdfc63efafe5b349051 # v5
+        uses: docker/metadata-action@dc802804100637a589fabce1cb79ff13a1411302 # v6.2.0
        with:
          images: |
            calciumion/new-api
      - name: Build & push single-arch
        id: build
-        uses: docker/build-push-action@10e90e3645eae34f1e60eeb005ba3a3d33f178e8 # v6
+        uses: docker/build-push-action@53b7df96c91f9c12dcc8a07bcb9ccacbed38856a # v7.3.0
        with:
          context: .
          platforms: ${{ matrix.platform }}
@@ -111,7 +111,7 @@ jobs:
          sbom: true
      - name: Install cosign
-        uses: sigstore/cosign-installer@398d4b0eeef1380460a10c8013a76f728fb906ac # v3
+        uses: sigstore/cosign-installer@6f9f17788090df1f26f669e9d70d6ae9567deba6 # v4.1.2
      - name: Sign image with cosign
        run: cosign sign --yes calciumion/new-api@${{ steps.build.outputs.digest }}
@@ -133,7 +133,7 @@ jobs:
      id-token: write
    steps:
      - name: Log in to Docker Hub
-        uses: docker/login-action@c94ce9fb468520275223c153574b00df6fe4bcc9 # v3
+        uses: docker/login-action@c99871dec2022cc055c062a10cc1a1310835ceb4 # v4.3.0
        with:
          username: ${{ secrets.DOCKERHUB_USERNAME }}
          password: ${{ secrets.DOCKERHUB_TOKEN }}
@@ -153,7 +153,7 @@ jobs:
            calciumion/new-api:${{ needs.prepare.outputs.version }}-arm64
      - name: Install cosign
-        uses: sigstore/cosign-installer@398d4b0eeef1380460a10c8013a76f728fb906ac # v3
+        uses: sigstore/cosign-installer@6f9f17788090df1f26f669e9d70d6ae9567deba6 # v4.1.2
      - name: Sign manifests with cosign
        run: |

--- a/.github/workflows/electron-build.yml
+++ b/.github/workflows/electron-build.yml
@@ -22,22 +22,22 @@ jobs:
    steps:
      - name: Checkout code
-        uses: actions/checkout@v4
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
        with:
          fetch-depth: 0
      - name: Setup Bun
-        uses: oven-sh/setup-bun@v2
+        uses: oven-sh/setup-bun@0c5077e51419868618aeaa5fe8019c62421857d6 # v2.2.0
        with:
          bun-version: latest
      - name: Setup Node.js
-        uses: actions/setup-node@v4
+        uses: actions/setup-node@48b55a011bda9f5d6aeb4c2d9c7362e8dae4041e # v6.4.0
        with:
          node-version: '20'
      - name: Setup Go
-        uses: actions/setup-go@v5
+        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
        with:
          go-version: '>=1.25.1'
@@ -106,7 +106,7 @@ jobs:
      # - name: Upload artifacts (macOS)
      #   if: runner.os == 'macOS'
-      #   uses: actions/upload-artifact@v4
+      #   uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
      #   with:
      #     name: macos-build
      #     path: |
@@ -115,7 +115,7 @@ jobs:
      - name: Upload artifacts (Windows)
        if: runner.os == 'Windows'
-        uses: actions/upload-artifact@v4
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
        with:
          name: windows-build
          path: |
@@ -130,12 +130,12 @@ jobs:
    steps:
      - name: Download all artifacts
-        uses: actions/download-artifact@v4
+        uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c # v8.0.1
      - name: Upload to Release
-        uses: softprops/action-gh-release@v2
+        uses: softprops/action-gh-release@718ea10b132b3b2eba29c1007bb80653f286566b # v3.0.1
        with:
          files: |
            windows-build/*
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
\ No newline at end of file
--- a/.github/workflows/pr-check.yml
+++ b/.github/workflows/pr-check.yml
@@ -13,7 +13,7 @@ jobs:
  pr-quality:
    runs-on: ubuntu-latest
    steps:
-      - uses: peakoss/anti-slop@v0.2.1
+      - uses: peakoss/anti-slop@85daca1880e9e1af197fc06ea03349daf08f4202 # v0.2.1
        with:
          max-failures: 4
          require-description: true

--- a/.github/workflows/release.yml
+++ b/.github/workflows/release.yml
@@ -19,14 +19,14 @@ jobs:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout
-        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
        with:
          fetch-depth: 0
      - name: Determine Version
        run: |
          VERSION=$(git describe --tags)
          echo "VERSION=$VERSION" >> $GITHUB_ENV
-      - uses: oven-sh/setup-bun@0c5077e51419868618aeaa5fe8019c62421857d6 # v2
+      - uses: oven-sh/setup-bun@0c5077e51419868618aeaa5fe8019c62421857d6 # v2.2.0
        with:
          bun-version: latest
      - name: Build Frontend (default)
@@ -48,7 +48,7 @@ jobs:
          VITE_REACT_APP_VERSION=$VERSION bun run build
          cd ../..
      - name: Set up Go
-        uses: actions/setup-go@40f1582b2485089dde7abd97c1529aa768e1baff # v5
+        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
        with:
          go-version: '>=1.25.1'
      - name: Build Backend (amd64)
@@ -64,7 +64,7 @@ jobs:
        run: sha256sum new-api-* > checksums-linux.txt
      - name: Release
-        uses: softprops/action-gh-release@153bb8e04406b158c6c84fc1615b65b24149a1fe # v2
+        uses: softprops/action-gh-release@718ea10b132b3b2eba29c1007bb80653f286566b # v3.0.1
        if: startsWith(github.ref, 'refs/tags/')
        with:
          files: |
@@ -78,14 +78,14 @@ jobs:
    runs-on: macos-latest
    steps:
      - name: Checkout
-        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
        with:
          fetch-depth: 0
      - name: Determine Version
        run: |
          VERSION=$(git describe --tags)
          echo "VERSION=$VERSION" >> $GITHUB_ENV
-      - uses: oven-sh/setup-bun@0c5077e51419868618aeaa5fe8019c62421857d6 # v2
+      - uses: oven-sh/setup-bun@0c5077e51419868618aeaa5fe8019c62421857d6 # v2.2.0
        with:
          bun-version: latest
      - name: Build Frontend (default)
@@ -108,7 +108,7 @@ jobs:
          VITE_REACT_APP_VERSION=$VERSION bun run build
          cd ../..
      - name: Set up Go
-        uses: actions/setup-go@40f1582b2485089dde7abd97c1529aa768e1baff # v5
+        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
        with:
          go-version: '>=1.25.1'
      - name: Build Backend
@@ -119,7 +119,7 @@ jobs:
        run: shasum -a 256 new-api-macos-* > checksums-macos.txt
      - name: Release
-        uses: softprops/action-gh-release@153bb8e04406b158c6c84fc1615b65b24149a1fe # v2
+        uses: softprops/action-gh-release@718ea10b132b3b2eba29c1007bb80653f286566b # v3.0.1
        if: startsWith(github.ref, 'refs/tags/')
        with:
          files: |
@@ -136,14 +136,14 @@ jobs:
        shell: bash
    steps:
      - name: Checkout
-        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
        with:
          fetch-depth: 0
      - name: Determine Version
        run: |
          VERSION=$(git describe --tags)
          echo "VERSION=$VERSION" >> $GITHUB_ENV
-      - uses: oven-sh/setup-bun@0c5077e51419868618aeaa5fe8019c62421857d6 # v2
+      - uses: oven-sh/setup-bun@0c5077e51419868618aeaa5fe8019c62421857d6 # v2.2.0
        with:
          bun-version: latest
      - name: Build Frontend (default)
@@ -165,7 +165,7 @@ jobs:
          VITE_REACT_APP_VERSION=$VERSION bun run build
          cd ../..
      - name: Set up Go
-        uses: actions/setup-go@40f1582b2485089dde7abd97c1529aa768e1baff # v5
+        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
        with:
          go-version: '>=1.25.1'
      - name: Build Backend
@@ -176,7 +176,7 @@ jobs:
        run: sha256sum new-api-*.exe > checksums-windows.txt
      - name: Release
-        uses: softprops/action-gh-release@153bb8e04406b158c6c84fc1615b65b24149a1fe # v2
+        uses: softprops/action-gh-release@718ea10b132b3b2eba29c1007bb80653f286566b # v3.0.1
        if: startsWith(github.ref, 'refs/tags/')
        with:
          files: |

--- a/controller/model_list_test.go
+++ b/controller/model_list_test.go
@@ -131,7 +131,17 @@ func withSelfUseModeDisabled(t *testing.T) {
 	})
 }
-func decodeListModelsResponse(t *testing.T, recorder *httptest.ResponseRecorder) map[string]struct{} {
+func withSelfUseModeEnabled(t *testing.T) {
+	t.Helper()
+	original := operation_setting.SelfUseModeEnabled
+	operation_setting.SelfUseModeEnabled = true
+	t.Cleanup(func() {
+		operation_setting.SelfUseModeEnabled = original
+	})
+}
+func decodeListModelsPayload(t *testing.T, recorder *httptest.ResponseRecorder) listModelsResponse {
 	t.Helper()
 	require.Equal(t, http.StatusOK, recorder.Code)
@@ -139,7 +149,13 @@ func decodeListModelsResponse(t *testing.T, recorder *httptest.ResponseRecorder)
 	require.NoError(t, common.Unmarshal(recorder.Body.Bytes(), &payload))
 	require.True(t, payload.Success)
 	require.Equal(t, "list", payload.Object)
+	return payload
+}
+func decodeListModelsResponse(t *testing.T, recorder *httptest.ResponseRecorder) map[string]struct{} {
+	t.Helper()
+	payload := decodeListModelsPayload(t, recorder)
 	ids := make(map[string]struct{}, len(payload.Data))
 	for _, item := range payload.Data {
 		ids[item.Id] = struct{}{}
@@ -255,6 +271,77 @@ func TestListModelsIncludesTieredBillingModel(t *testing.T) {
 	require.Empty(t, missingExprPricing.BillingExpr)
 }
+func TestListModelsUsesAdvancedCustomEndpointTypesFromPricingCache(t *testing.T) {
+	withSelfUseModeEnabled(t)
+	db := setupModelListControllerTestDB(t)
+	originalMemoryCacheEnabled := common.MemoryCacheEnabled
+	common.MemoryCacheEnabled = true
+	t.Cleanup(func() {
+		common.MemoryCacheEnabled = originalMemoryCacheEnabled
+		model.InvalidatePricingCache()
+	})
+	require.NoError(t, db.Create(&model.User{
+		Id:       1003,
+		Username: "advanced-custom-model-list-user",
+		Password: "password",
+		Group:    "default",
+		Status:   common.UserStatusEnabled,
+	}).Error)
+	channel := &model.Channel{
+		Id:     701,
+		Type:   constant.ChannelTypeAdvancedCustom,
+		Key:    "advanced-custom-key",
+		Status: common.ChannelStatusEnabled,
+		Name:   "advanced-custom-channel",
+		Group:  "default",
+		Models: "gemini-3.5-flash",
+	}
+	channel.SetOtherSettings(dto.ChannelOtherSettings{
+		AdvancedCustom: &dto.AdvancedCustomConfig{
+			Routes: []dto.AdvancedCustomRoute{
+				{
+					IncomingPath: "/v1/chat/completions",
+					UpstreamPath: "/v1/chat/completions",
+				},
+				{
+					IncomingPath: "/v1/responses",
+					UpstreamPath: "/v1beta/models/{model}:generateContent",
+					Converter:    "openai_responses_to_gemini_generate_content",
+					Models:       []string{"re:^gemini-"},
+				},
+			},
+		},
+	})
+	require.NoError(t, db.Create(channel).Error)
+	require.NoError(t, db.Create(&model.Ability{
+		Group:     "default",
+		Model:     "gemini-3.5-flash",
+		ChannelId: 701,
+		Enabled:   true,
+	}).Error)
+	model.InitChannelCache()
+	model.GetPricing()
+	recorder := httptest.NewRecorder()
+	ctx, _ := gin.CreateTestContext(recorder)
+	ctx.Request = httptest.NewRequest(http.MethodGet, "/v1/models", nil)
+	ctx.Set("id", 1003)
+	ListModels(ctx, constant.ChannelTypeOpenAI)
+	payload := decodeListModelsPayload(t, recorder)
+	require.Len(t, payload.Data, 1)
+	require.Equal(t, "gemini-3.5-flash", payload.Data[0].Id)
+	require.Equal(t, []constant.EndpointType{
+		constant.EndpointTypeOpenAI,
+		constant.EndpointTypeOpenAIResponse,
+	}, payload.Data[0].SupportedEndpointTypes)
+}
 func TestListModelsTokenLimitIncludesTieredBillingModel(t *testing.T) {
 	withSelfUseModeDisabled(t)
 	withTieredBillingConfig(t, map[string]string{

--- a/dto/billing_usage.go
+++ b/dto/billing_usage.go
+package dto
+const (
+	BillingUsageSourceClaudeMessages = "claude_messages"
+	BillingUsageSourceGeminiChat     = "gemini_chat"
+	BillingUsageSourceOAIChat        = "oai_chat"
+	BillingUsageSourceOAIResponses   = "oai_responses"
+	BillingUsageSemanticAnthropic = "anthropic"
+	BillingUsageSemanticGemini    = "gemini"
+	BillingUsageSemanticOpenAI    = "openai"
+)
+type BillingUsage struct {
+	Source              string               `json:"source,omitempty"`
+	Semantic            string               `json:"semantic,omitempty"`
+	Estimated           bool                 `json:"estimated,omitempty"`
+	OpenAIUsage         *Usage               `json:"openai_usage,omitempty"`
+	ClaudeUsage         *ClaudeUsage         `json:"claude_usage,omitempty"`
+	GeminiUsageMetadata *GeminiUsageMetadata `json:"gemini_usage_metadata,omitempty"`
+}
+func NewClaudeMessagesBillingUsage(usage *ClaudeUsage) *BillingUsage {
+	if !HasClaudeUsageTokens(usage) {
+		return nil
+	}
+	return &BillingUsage{
+		Source:      BillingUsageSourceClaudeMessages,
+		Semantic:    BillingUsageSemanticAnthropic,
+		ClaudeUsage: cloneClaudeUsage(usage),
+	}
+}
+// HasClaudeUsageTokens mirrors HasOpenAIUsageTokens/HasGeminiUsageMetadataTokens:
+// an all-zero ClaudeUsage must not become a BillingUsage, otherwise it would take
+// precedence during settlement and zero out a non-zero top-level usage.
+func HasClaudeUsageTokens(usage *ClaudeUsage) bool {
+	if usage == nil {
+		return false
+	}
+	if usage.InputTokens != 0 ||
+		usage.OutputTokens != 0 ||
+		usage.CacheCreationInputTokens != 0 ||
+		usage.CacheReadInputTokens != 0 ||
+		usage.ClaudeCacheCreation5mTokens != 0 ||
+		usage.ClaudeCacheCreation1hTokens != 0 {
+		return true
+	}
+	if usage.CacheCreation != nil &&
+		(usage.CacheCreation.Ephemeral5mInputTokens != 0 || usage.CacheCreation.Ephemeral1hInputTokens != 0) {
+		return true
+	}
+	return false
+}
+func NewOpenAIChatBillingUsage(usage *Usage) *BillingUsage {
+	return newOpenAIBillingUsage(BillingUsageSourceOAIChat, usage)
+}
+func NewOpenAIResponsesBillingUsage(usage *Usage) *BillingUsage {
+	return newOpenAIBillingUsage(BillingUsageSourceOAIResponses, usage)
+}
+func newOpenAIBillingUsage(source string, usage *Usage) *BillingUsage {
+	if !HasOpenAIUsageTokens(usage) {
+		return nil
+	}
+	return &BillingUsage{
+		Source:      source,
+		Semantic:    BillingUsageSemanticOpenAI,
+		OpenAIUsage: cloneOpenAIUsage(usage),
+	}
+}
+func HasOpenAIUsageTokens(usage *Usage) bool {
+	if usage == nil {
+		return false
+	}
+	if usage.PromptTokens != 0 ||
+		usage.CompletionTokens != 0 ||
+		usage.TotalTokens != 0 ||
+		usage.InputTokens != 0 ||
+		usage.OutputTokens != 0 ||
+		usage.PromptCacheHitTokens != 0 ||
+		usage.ClaudeCacheCreation5mTokens != 0 ||
+		usage.ClaudeCacheCreation1hTokens != 0 {
+		return true
+	}
+	if usage.PromptTokensDetails.CachedTokens != 0 ||
+		usage.PromptTokensDetails.CachedCreationTokens != 0 ||
+		usage.PromptTokensDetails.TextTokens != 0 ||
+		usage.PromptTokensDetails.ImageTokens != 0 ||
+		usage.PromptTokensDetails.AudioTokens != 0 {
+		return true
+	}
+	if usage.CompletionTokenDetails.ReasoningTokens != 0 ||
+		usage.CompletionTokenDetails.TextTokens != 0 ||
+		usage.CompletionTokenDetails.ImageTokens != 0 ||
+		usage.CompletionTokenDetails.AudioTokens != 0 {
+		return true
+	}
+	return usage.InputTokensDetails != nil
+}
+func NewGeminiChatBillingUsage(metadata *GeminiUsageMetadata) *BillingUsage {
+	return newGeminiChatBillingUsage(metadata, false)
+}
+func NewEstimatedGeminiChatBillingUsage(usage *Usage) *BillingUsage {
+	if usage == nil {
+		return nil
+	}
+	totalTokens := usage.TotalTokens
+	if totalTokens == 0 {
+		totalTokens = usage.PromptTokens + usage.CompletionTokens
+	}
+	return newGeminiChatBillingUsage(&GeminiUsageMetadata{
+		PromptTokenCount:     usage.PromptTokens,
+		CandidatesTokenCount: usage.CompletionTokens,
+		TotalTokenCount:      totalTokens,
+	}, true)
+}
+func newGeminiChatBillingUsage(metadata *GeminiUsageMetadata, estimated bool) *BillingUsage {
+	if !HasGeminiUsageMetadataTokens(metadata) {
+		return nil
+	}
+	usageMetadata := cloneGeminiUsageMetadata(*metadata)
+	return &BillingUsage{
+		Source:              BillingUsageSourceGeminiChat,
+		Semantic:            BillingUsageSemanticGemini,
+		Estimated:           estimated,
+		GeminiUsageMetadata: &usageMetadata,
+	}
+}
+func CloneBillingUsage(usage *BillingUsage) *BillingUsage {
+	if usage == nil {
+		return nil
+	}
+	clone := *usage
+	clone.OpenAIUsage = cloneOpenAIUsage(usage.OpenAIUsage)
+	clone.ClaudeUsage = cloneClaudeUsage(usage.ClaudeUsage)
+	if usage.GeminiUsageMetadata != nil {
+		metadata := cloneGeminiUsageMetadata(*usage.GeminiUsageMetadata)
+		clone.GeminiUsageMetadata = &metadata
+	}
+	return &clone
+}
+func cloneOpenAIUsage(usage *Usage) *Usage {
+	if usage == nil {
+		return nil
+	}
+	clone := *usage
+	clone.BillingUsage = nil
+	if usage.InputTokensDetails != nil {
+		inputTokensDetails := *usage.InputTokensDetails
+		clone.InputTokensDetails = &inputTokensDetails
+	}
+	return &clone
+}
+func cloneClaudeUsage(usage *ClaudeUsage) *ClaudeUsage {
+	if usage == nil {
+		return nil
+	}
+	clone := *usage
+	clone.BillingUsage = nil
+	if usage.CacheCreation != nil {
+		cacheCreation := *usage.CacheCreation
+		clone.CacheCreation = &cacheCreation
+	}
+	if usage.ServerToolUse != nil {
+		serverToolUse := *usage.ServerToolUse
+		clone.ServerToolUse = &serverToolUse
+	}
+	return &clone
+}
+func cloneGeminiUsageMetadata(metadata GeminiUsageMetadata) GeminiUsageMetadata {
+	metadata.PromptTokensDetails = append([]GeminiPromptTokensDetails{}, metadata.PromptTokensDetails...)
+	metadata.ToolUsePromptTokensDetails = append([]GeminiPromptTokensDetails{}, metadata.ToolUsePromptTokensDetails...)
+	metadata.CandidatesTokensDetails = append([]GeminiPromptTokensDetails{}, metadata.CandidatesTokensDetails...)
+	metadata.BillingUsage = nil
+	return metadata
+}
+func HasGeminiUsageMetadataTokens(metadata *GeminiUsageMetadata) bool {
+	if metadata == nil {
+		return false
+	}
+	if metadata.PromptTokenCount != 0 ||
+		metadata.ToolUsePromptTokenCount != 0 ||
+		metadata.CandidatesTokenCount != 0 ||
+		metadata.TotalTokenCount != 0 ||
+		metadata.ThoughtsTokenCount != 0 ||
+		metadata.CachedContentTokenCount != 0 {
+		return true
+	}
+	for _, detail := range metadata.PromptTokensDetails {
+		if detail.TokenCount != 0 {
+			return true
+		}
+	}
+	for _, detail := range metadata.ToolUsePromptTokensDetails {
+		if detail.TokenCount != 0 {
+			return true
+		}
+	}
+	for _, detail := range metadata.CandidatesTokensDetails {
+		if detail.TokenCount != 0 {
+			return true
+		}
+	}
+	return false
+}
--- a/dto/billing_usage_test.go
+++ b/dto/billing_usage_test.go
+package dto
+import (
+	"testing"
+	"github.com/QuantumNous/new-api/common"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+func TestNewGeminiChatBillingUsageRequiresTokenContent(t *testing.T) {
+	require.Nil(t, NewGeminiChatBillingUsage(nil))
+	require.Nil(t, NewGeminiChatBillingUsage(&GeminiUsageMetadata{}))
+	billingUsage := NewGeminiChatBillingUsage(&GeminiUsageMetadata{PromptTokenCount: 1})
+	require.NotNil(t, billingUsage)
+	require.NotNil(t, billingUsage.GeminiUsageMetadata)
+	assert.Equal(t, BillingUsageSourceGeminiChat, billingUsage.Source)
+	assert.Equal(t, BillingUsageSemanticGemini, billingUsage.Semantic)
+	assert.False(t, billingUsage.Estimated)
+}
+func TestNewClaudeMessagesBillingUsageRequiresTokenContent(t *testing.T) {
+	require.Nil(t, NewClaudeMessagesBillingUsage(nil))
+	require.Nil(t, NewClaudeMessagesBillingUsage(&ClaudeUsage{}))
+	require.Nil(t, NewClaudeMessagesBillingUsage(&ClaudeUsage{CacheCreation: &ClaudeCacheCreationUsage{}}))
+	billingUsage := NewClaudeMessagesBillingUsage(&ClaudeUsage{InputTokens: 1})
+	require.NotNil(t, billingUsage)
+	require.NotNil(t, billingUsage.ClaudeUsage)
+	assert.Equal(t, BillingUsageSourceClaudeMessages, billingUsage.Source)
+	assert.Equal(t, BillingUsageSemanticAnthropic, billingUsage.Semantic)
+	cacheOnly := NewClaudeMessagesBillingUsage(&ClaudeUsage{
+		CacheCreation: &ClaudeCacheCreationUsage{Ephemeral5mInputTokens: 4},
+	})
+	require.NotNil(t, cacheOnly)
+}
+func TestNewOpenAIChatBillingUsageRequiresTokenContent(t *testing.T) {
+	require.Nil(t, NewOpenAIChatBillingUsage(nil))
+	require.Nil(t, NewOpenAIChatBillingUsage(&Usage{}))
+	billingUsage := NewOpenAIChatBillingUsage(&Usage{PromptTokens: 1})
+	require.NotNil(t, billingUsage)
+	require.NotNil(t, billingUsage.OpenAIUsage)
+	assert.Equal(t, BillingUsageSourceOAIChat, billingUsage.Source)
+	assert.Equal(t, BillingUsageSemanticOpenAI, billingUsage.Semantic)
+	assert.Equal(t, 1, billingUsage.OpenAIUsage.PromptTokens)
+}
+func TestNewEstimatedGeminiChatBillingUsage(t *testing.T) {
+	billingUsage := NewEstimatedGeminiChatBillingUsage(&Usage{
+		PromptTokens:     11,
+		CompletionTokens: 7,
+	})
+	require.NotNil(t, billingUsage)
+	require.NotNil(t, billingUsage.GeminiUsageMetadata)
+	assert.True(t, billingUsage.Estimated)
+	assert.Equal(t, 11, billingUsage.GeminiUsageMetadata.PromptTokenCount)
+	assert.Equal(t, 7, billingUsage.GeminiUsageMetadata.CandidatesTokenCount)
+	assert.Equal(t, 18, billingUsage.GeminiUsageMetadata.TotalTokenCount)
+}
+func TestBillingUsageJSONUsesProtocolNamedFields(t *testing.T) {
+	billingUsage := &BillingUsage{
+		OpenAIUsage:         &Usage{PromptTokens: 1, BillingUsage: NewClaudeMessagesBillingUsage(&ClaudeUsage{InputTokens: 9})},
+		ClaudeUsage:         &ClaudeUsage{InputTokens: 2, BillingUsage: NewOpenAIChatBillingUsage(&Usage{PromptTokens: 8})},
+		GeminiUsageMetadata: &GeminiUsageMetadata{PromptTokenCount: 3, BillingUsage: NewOpenAIChatBillingUsage(&Usage{PromptTokens: 7})},
+	}
+	data, err := common.Marshal(billingUsage)
+	require.NoError(t, err)
+	assert.Contains(t, string(data), `"openai_usage"`)
+	assert.Contains(t, string(data), `"claude_usage"`)
+	assert.Contains(t, string(data), `"gemini_usage_metadata"`)
+	assert.NotContains(t, string(data), `"usage":`)
+	assert.NotContains(t, string(data), `"usage_metadata"`)
+	clone := CloneBillingUsage(billingUsage)
+	require.NotNil(t, clone.OpenAIUsage)
+	require.NotNil(t, clone.ClaudeUsage)
+	require.NotNil(t, clone.GeminiUsageMetadata)
+	assert.Nil(t, clone.OpenAIUsage.BillingUsage)
+	assert.Nil(t, clone.ClaudeUsage.BillingUsage)
+	assert.Nil(t, clone.GeminiUsageMetadata.BillingUsage)
+}
--- a/dto/channel_settings.go
+++ b/dto/channel_settings.go
--- a/dto/channel_settings_test.go
+++ b/dto/channel_settings_test.go
--- a/dto/claude.go
+++ b/dto/claude.go
@@ -564,6 +564,7 @@ type ClaudeUsage struct {
 	ClaudeCacheCreation5mTokens int                  `json:"claude_cache_creation_5_m_tokens"`
 	ClaudeCacheCreation1hTokens int                  `json:"claude_cache_creation_1_h_tokens"`
 	ServerToolUse               *ClaudeServerToolUse `json:"server_tool_use,omitempty"`
+	BillingUsage                *BillingUsage        `json:"billing_usage,omitempty"`
 }
 type ClaudeCacheCreationUsage struct {

--- a/dto/gemini.go
+++ b/dto/gemini.go
@@ -44,9 +44,9 @@ func (r *GeminiChatRequest) UnmarshalJSON(data []byte) error {
 }
 type ToolConfig struct {
-	FunctionCallingConfig *FunctionCallingConfig `json:"functionCallingConfig,omitempty"`
+	FunctionCallingConfig            *FunctionCallingConfig `json:"functionCallingConfig,omitempty"`
-	RetrievalConfig       *RetrievalConfig       `json:"retrievalConfig,omitempty"`
+	RetrievalConfig                  *RetrievalConfig       `json:"retrievalConfig,omitempty"`
-	IncludeServerSideToolInvocations *bool       `json:"includeServerSideToolInvocations,omitempty"`
+	IncludeServerSideToolInvocations *bool                  `json:"includeServerSideToolInvocations,omitempty"`
 }
 type FunctionCallingConfig struct {
@@ -455,9 +455,46 @@ type GeminiChatPromptFeedback struct {
 }
 type GeminiChatResponse struct {
-	Candidates     []GeminiChatCandidate     `json:"candidates"`
+	Candidates       []GeminiChatCandidate     `json:"candidates"`
-	PromptFeedback *GeminiChatPromptFeedback `json:"promptFeedback,omitempty"`
+	PromptFeedback   *GeminiChatPromptFeedback `json:"promptFeedback,omitempty"`
-	UsageMetadata  GeminiUsageMetadata       `json:"usageMetadata"`
+	UsageMetadata    GeminiUsageMetadata       `json:"usageMetadata"`
+	HasUsageMetadata bool                      `json:"-"`
+}
+// UnmarshalJSON records whether Gemini returned usageMetadata while preserving
+// the historical wire shape that always marshals the usageMetadata field.
+//
+// IMPORTANT: aux shadows GeminiChatResponse. Any field added to
+// GeminiChatResponse must also be added to aux (and copied below), otherwise it
+// is silently dropped during unmarshal.
+func (r *GeminiChatResponse) UnmarshalJSON(data []byte) error {
+	var aux struct {
+		Candidates     []GeminiChatCandidate     `json:"candidates"`
+		PromptFeedback *GeminiChatPromptFeedback `json:"promptFeedback,omitempty"`
+		UsageMetadata  *GeminiUsageMetadata      `json:"usageMetadata"`
+	}
+	if err := common.Unmarshal(data, &aux); err != nil {
+		return err
+	}
+	r.Candidates = aux.Candidates
+	r.PromptFeedback = aux.PromptFeedback
+	r.HasUsageMetadata = aux.UsageMetadata != nil
+	if aux.UsageMetadata != nil {
+		r.UsageMetadata = *aux.UsageMetadata
+	} else {
+		r.UsageMetadata = GeminiUsageMetadata{}
+	}
+	return nil
+}
+func (r *GeminiChatResponse) GetUsageMetadata() *GeminiUsageMetadata {
+	if r == nil {
+		return nil
+	}
+	if r.HasUsageMetadata || HasGeminiUsageMetadataTokens(&r.UsageMetadata) {
+		return &r.UsageMetadata
+	}
+	return nil
 }
 type GeminiUsageMetadata struct {
@@ -470,6 +507,7 @@ type GeminiUsageMetadata struct {
 	PromptTokensDetails        []GeminiPromptTokensDetails `json:"promptTokensDetails"`
 	ToolUsePromptTokensDetails []GeminiPromptTokensDetails `json:"toolUsePromptTokensDetails"`
 	CandidatesTokensDetails    []GeminiPromptTokensDetails `json:"candidatesTokensDetails"`
+	BillingUsage               *BillingUsage               `json:"billing_usage,omitempty"`
 }
 type GeminiPromptTokensDetails struct {

--- a/dto/gemini_response_test.go
+++ b/dto/gemini_response_test.go
+package dto
+import (
+	"testing"
+	"github.com/QuantumNous/new-api/common"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+func TestGeminiChatResponseUsageMetadataPresence(t *testing.T) {
+	var missing GeminiChatResponse
+	require.NoError(t, common.Unmarshal([]byte(`{"candidates":[]}`), &missing))
+	assert.False(t, missing.HasUsageMetadata)
+	assert.Nil(t, missing.GetUsageMetadata())
+	var empty GeminiChatResponse
+	require.NoError(t, common.Unmarshal([]byte(`{"candidates":[],"usageMetadata":{}}`), &empty))
+	assert.True(t, empty.HasUsageMetadata)
+	require.NotNil(t, empty.GetUsageMetadata())
+	assert.False(t, HasGeminiUsageMetadataTokens(empty.GetUsageMetadata()))
+	var populated GeminiChatResponse
+	require.NoError(t, common.Unmarshal([]byte(`{"candidates":[],"usageMetadata":{"promptTokenCount":3}}`), &populated))
+	assert.True(t, populated.HasUsageMetadata)
+	require.NotNil(t, populated.GetUsageMetadata())
+	assert.True(t, HasGeminiUsageMetadataTokens(populated.GetUsageMetadata()))
+}
+func TestGeminiChatResponseMarshalKeepsUsageMetadataField(t *testing.T) {
+	data, err := common.Marshal(GeminiChatResponse{})
+	require.NoError(t, err)
+	assert.Contains(t, string(data), `"usageMetadata"`)
+}
--- a/dto/openai_response.go
+++ b/dto/openai_response.go
@@ -221,12 +221,13 @@ type CompletionsStreamResponse struct {
 }
 type Usage struct {
-	PromptTokens         int    `json:"prompt_tokens"`
+	PromptTokens         int           `json:"prompt_tokens"`
-	CompletionTokens     int    `json:"completion_tokens"`
+	CompletionTokens     int           `json:"completion_tokens"`
-	TotalTokens          int    `json:"total_tokens"`
+	TotalTokens          int           `json:"total_tokens"`
-	PromptCacheHitTokens int    `json:"prompt_cache_hit_tokens,omitempty"`
+	PromptCacheHitTokens int           `json:"prompt_cache_hit_tokens,omitempty"`
-	UsageSemantic        string `json:"usage_semantic,omitempty"`
+	UsageSemantic        string        `json:"usage_semantic,omitempty"`
-	UsageSource          string `json:"usage_source,omitempty"`
+	UsageSource          string        `json:"usage_source,omitempty"`
+	BillingUsage         *BillingUsage `json:"billing_usage,omitempty"`
 	PromptTokensDetails    InputTokenDetails  `json:"prompt_tokens_details"`
 	CompletionTokenDetails OutputTokenDetails `json:"completion_tokens_details"`

--- a/main.go
+++ b/main.go
@@ -102,6 +102,10 @@ func main() {
 		go model.SyncChannelCache(common.SyncFrequency)
 	}
+	// Warm pricing after channel cache initialization so Advanced Custom
+	// endpoint inference can read cached route settings on first request.
+	model.GetPricing()
 	// 热更新配置
 	go model.SyncOptions(common.SyncFrequency)
@@ -330,9 +334,6 @@ func InitResources() error {
 	// 清理旧的磁盘缓存文件
 	common.CleanupOldCacheFiles()
-	// 初始化模型
-	model.GetPricing()
 	// Initialize SQL Database
 	err = model.InitLogDB()
 	if err != nil {

--- a/middleware/distributor.go
+++ b/middleware/distributor.go
@@ -105,7 +105,7 @@ func Distribute() func(c *gin.Context) {
 					affinityUsable := false
 					preferred, err := model.CacheGetChannel(preferredChannelID)
 					if err == nil && preferred != nil && preferred.Status == common.ChannelStatusEnabled &&
-						channelSupportsRequestPath(preferred, c.Request.URL.Path) {
+						channelSupportsRequestPath(preferred, c.Request.URL.Path, modelRequest.Model) {
 						if usingGroup == "auto" {
 							userGroup := common.GetContextKeyString(c, constant.ContextKeyUserGroup)
 							autoGroups := service.GetUserAutoGroup(userGroup)
@@ -172,7 +172,7 @@ func Distribute() func(c *gin.Context) {
 // channelSupportsRequestPath reports whether a channel can serve the request path.
 // Only Advanced Custom (type 58) channels are path-checked; all other channel types
 // always pass. A type-58 channel is usable only when one of its routes matches.
-func channelSupportsRequestPath(channel *model.Channel, requestPath string) bool {
+func channelSupportsRequestPath(channel *model.Channel, requestPath string, requestModel string) bool {
 	if channel == nil {
 		return false
 	}
@@ -180,7 +180,7 @@ func channelSupportsRequestPath(channel *model.Channel, requestPath string) bool
 		return true
 	}
 	config := channel.GetOtherSettings().AdvancedCustom
-	return config != nil && config.SupportsPath(requestPath)
+	return config != nil && config.SupportsPathForModel(requestPath, requestModel)
 }
 // getModelFromRequest 从请求中读取模型信息

--- a/model/ability.go
+++ b/model/ability.go
@@ -121,7 +121,7 @@ func GetChannel(group string, model string, retry int, requestPath string) (*Cha
 	if err != nil {
 		return nil, err
 	}
-	abilities = filterAbilitiesByRequestPath(abilities, requestPath)
+	abilities = filterAbilitiesByRequestPathAndModel(abilities, requestPath, model)
 	channel := Channel{}
 	if len(abilities) > 0 {
 		// Randomly choose one
@@ -146,11 +146,12 @@ func GetChannel(group string, model string, retry int, requestPath string) (*Cha
 	return &channel, err
 }
-// filterAbilitiesByRequestPath restricts candidates by request path for the DB
+// filterAbilitiesByRequestPathAndModel restricts candidates by request path and
-// (non-memory-cache) selection path. Only Advanced Custom (type 58) channels are
+// model for the DB (non-memory-cache) selection path. Only Advanced Custom
-// path-checked: kept only when one of their routes matches requestPath; all other
+// (type 58) channels are path-checked: kept only when one of their routes matches
-// channel types always pass. When requestPath is empty, filtering is skipped.
+// requestPath and model; all other channel types always pass. When requestPath is
-func filterAbilitiesByRequestPath(abilities []Ability, requestPath string) []Ability {
+// empty, filtering is skipped.
+func filterAbilitiesByRequestPathAndModel(abilities []Ability, requestPath string, model string) []Ability {
 	if requestPath == "" || len(abilities) == 0 {
 		return abilities
 	}
@@ -185,7 +186,7 @@ func filterAbilitiesByRequestPath(abilities []Ability, requestPath string) []Abi
 			filtered = append(filtered, ability)
 			continue
 		}
-		if config != nil && config.SupportsPath(requestPath) {
+		if config != nil && config.SupportsPathForModel(requestPath, model) {
 			filtered = append(filtered, ability)
 		}
 	}

--- a/model/channel_cache.go
+++ b/model/channel_cache.go
@@ -25,6 +25,7 @@ var channelSyncLock sync.RWMutex
 func InitChannelCache() {
 	if !common.MemoryCacheEnabled {
+		InvalidatePricingCache()
 		return
 	}
 	newChannelId2channel := make(map[int]*Channel)
@@ -94,6 +95,11 @@ func InitChannelCache() {
 	channelsIDM = newChannelId2channel
 	channel2advancedCustomConfig = newChannel2advancedCustomConfig
 	channelSyncLock.Unlock()
+	// Lock ordering: InvalidatePricingCache acquires updatePricingLock, and
+	// GetPricing (holding updatePricingLock) nests channelSyncLock.RLock via
+	// loadPricingAdvancedCustomConfigs. channelSyncLock MUST be released before
+	// invalidating the pricing cache, otherwise the reversed order deadlocks.
+	InvalidatePricingCache()
 	common.SysLog("channels synced from database")
 }
@@ -115,12 +121,12 @@ func GetRandomSatisfiedChannel(group string, model string, retry int, requestPat
 	defer channelSyncLock.RUnlock()
 	// First, try to find channels with the exact model name.
-	channels := filterChannelsByRequestPath(group2model2channels[group][model], requestPath)
+	channels := filterChannelsByRequestPathAndModel(group2model2channels[group][model], requestPath, model)
 	// If no channels found, try to find channels with the normalized model name.
 	if len(channels) == 0 {
 		normalizedModel := ratio_setting.FormatMatchingModelName(model)
-		channels = filterChannelsByRequestPath(group2model2channels[group][normalizedModel], requestPath)
+		channels = filterChannelsByRequestPathAndModel(group2model2channels[group][normalizedModel], requestPath, model)
 	}
 	if len(channels) == 0 {
@@ -202,12 +208,12 @@ func GetRandomSatisfiedChannel(group string, model string, retry int, requestPat
 	return nil, errors.New("channel not found")
 }
-// filterChannelsByRequestPath restricts candidates by request path. Only Advanced
+// filterChannelsByRequestPathAndModel restricts candidates by request path and
-// Custom (type 58) channels are path-checked: they are kept only when one of their
+// model. Only Advanced Custom (type 58) channels are path-checked: they are kept
-// configured routes matches requestPath. All other channel types always pass.
+// only when one of their configured routes matches requestPath and model. All
-// When requestPath is empty (non-relay callers) filtering is skipped.
+// other channel types always pass. When requestPath is empty, filtering is skipped.
 // Caller must hold channelSyncLock (read lock). The cached slice is never mutated.
-func filterChannelsByRequestPath(channels []int, requestPath string) []int {
+func filterChannelsByRequestPathAndModel(channels []int, requestPath string, model string) []int {
 	if requestPath == "" || len(channels) == 0 {
 		return channels
 	}
@@ -223,7 +229,7 @@ func filterChannelsByRequestPath(channels []int, requestPath string) []int {
 			filtered = append(filtered, channelId)
 			continue
 		}
-		if config := channel2advancedCustomConfig[channelId]; config != nil && config.SupportsPath(requestPath) {
+		if config := channel2advancedCustomConfig[channelId]; config != nil && config.SupportsPathForModel(requestPath, model) {
 			filtered = append(filtered, channelId)
 		}
 	}
@@ -292,8 +298,8 @@ func CacheUpdateChannel(channel *Channel) {
 		return
 	}
 	channelSyncLock.Lock()
-	defer channelSyncLock.Unlock()
 	if channel == nil {
+		channelSyncLock.Unlock()
 		return
 	}
@@ -304,5 +310,20 @@ func CacheUpdateChannel(channel *Channel) {
 		logger.LogDebug(nil, "CacheUpdateChannel before: id=%d, name=%s, status=%d, polling_index=%d", channel.Id, channel.Name, channel.Status, oldChannel.ChannelInfo.MultiKeyPollingIndex)
 	}
 	channelsIDM[channel.Id] = channel
+	if channel2advancedCustomConfig == nil {
+		channel2advancedCustomConfig = make(map[int]*dto.AdvancedCustomConfig)
+	}
+	delete(channel2advancedCustomConfig, channel.Id)
+	if channel.Type == constant.ChannelTypeAdvancedCustom {
+		if config := channel.GetOtherSettings().AdvancedCustom; config != nil {
+			channel2advancedCustomConfig[channel.Id] = config
+		}
+	}
 	logger.LogDebug(nil, "CacheUpdateChannel after: id=%d, name=%s, status=%d, polling_index=%d", channel.Id, channel.Name, channel.Status, channel.ChannelInfo.MultiKeyPollingIndex)
+	// Lock ordering: do NOT hold channelSyncLock while calling
+	// InvalidatePricingCache. GetPricing acquires updatePricingLock first and then
+	// channelSyncLock.RLock (via loadPricingAdvancedCustomConfigs); acquiring
+	// updatePricingLock while holding channelSyncLock would be an AB-BA deadlock.
+	channelSyncLock.Unlock()
+	InvalidatePricingCache()
 }
--- a/model/pricing.go
+++ b/model/pricing.go
 package model
 import (
-	"encoding/json"
 	"fmt"
 	"strings"
@@ -10,6 +9,7 @@ import (
 	"github.com/QuantumNous/new-api/common"
 	"github.com/QuantumNous/new-api/constant"
+	"github.com/QuantumNous/new-api/dto"
 	"github.com/QuantumNous/new-api/setting/billing_setting"
 	"github.com/QuantumNous/new-api/setting/ratio_setting"
 	"github.com/QuantumNous/new-api/types"
@@ -107,6 +107,76 @@ func GetModelSupportEndpointTypes(model string) []constant.EndpointType {
 	return make([]constant.EndpointType, 0)
 }
+func getPricingEndpointTypesForAbility(ability AbilityWithChannel, advancedCustomConfigs map[int]*dto.AdvancedCustomConfig) []constant.EndpointType {
+	if ability.ChannelType != constant.ChannelTypeAdvancedCustom {
+		return common.GetEndpointTypesByChannelType(ability.ChannelType, ability.Model)
+	}
+	if config := advancedCustomConfigs[ability.ChannelId]; config != nil {
+		return config.SupportedEndpointTypesForModel(ability.Model)
+	}
+	return common.GetEndpointTypesByChannelType(ability.ChannelType, ability.Model)
+}
+// loadPricingAdvancedCustomConfigs runs inside updatePricing while
+// updatePricingLock is held, and nests channelSyncLock.RLock. This defines the
+// global lock order updatePricingLock -> channelSyncLock: any code path holding
+// channelSyncLock must release it before touching the pricing cache (see
+// InitChannelCache / CacheUpdateChannel), otherwise it deadlocks.
+// The returned configs are pointers shared with the channel cache; they are
+// replaced wholesale on update and never mutated in place, so reading them after
+// RUnlock is safe.
+func loadPricingAdvancedCustomConfigs(enableAbilities []AbilityWithChannel) map[int]*dto.AdvancedCustomConfig {
+	channelIDs := make([]int, 0)
+	seen := make(map[int]struct{})
+	for _, ability := range enableAbilities {
+		if ability.ChannelType != constant.ChannelTypeAdvancedCustom {
+			continue
+		}
+		if _, exists := seen[ability.ChannelId]; exists {
+			continue
+		}
+		seen[ability.ChannelId] = struct{}{}
+		channelIDs = append(channelIDs, ability.ChannelId)
+	}
+	if len(channelIDs) == 0 {
+		return nil
+	}
+	configs := make(map[int]*dto.AdvancedCustomConfig, len(channelIDs))
+	if common.MemoryCacheEnabled {
+		channelSyncLock.RLock()
+		defer channelSyncLock.RUnlock()
+		for _, channelID := range channelIDs {
+			if config := channel2advancedCustomConfig[channelID]; config != nil {
+				configs[channelID] = config
+			}
+		}
+		return configs
+	}
+	for _, channelID := range channelIDs {
+		channel, err := CacheGetChannel(channelID)
+		if err != nil {
+			common.SysLog(fmt.Sprintf("load advanced custom channel settings error: channel_id=%d, error=%v", channelID, err))
+			continue
+		}
+		if channel.Type != constant.ChannelTypeAdvancedCustom {
+			continue
+		}
+		if config := channel.GetOtherSettings().AdvancedCustom; config != nil {
+			configs[channelID] = config
+		}
+	}
+	return configs
+}
+func appendPricingEndpoint(endpoints []string, endpoint string) []string {
+	if endpoint == "" || common.StringsContains(endpoints, endpoint) {
+		return endpoints
+	}
+	return append(endpoints, endpoint)
+}
 func updatePricing() {
 	//modelRatios := common.GetModelRatios()
 	enableAbilities, err := GetAllEnableAbilityWithChannels()
@@ -201,11 +271,12 @@ func updatePricing() {
 	//这里使用切片而不是Set，因为一个模型可能支持多个端点类型，并且第一个端点是优先使用端点
 	modelSupportEndpointsStr := make(map[string][]string)
+	advancedCustomConfigs := loadPricingAdvancedCustomConfigs(enableAbilities)
 	// 先根据已有能力填充原生端点
 	for _, ability := range enableAbilities {
 		endpoints := modelSupportEndpointsStr[ability.Model]
-		channelTypes := common.GetEndpointTypesByChannelType(ability.ChannelType, ability.Model)
+		channelTypes := getPricingEndpointTypesForAbility(ability, advancedCustomConfigs)
 		for _, channelType := range channelTypes {
 			if !common.StringsContains(endpoints, string(channelType)) {
 				endpoints = append(endpoints, string(channelType))
@@ -214,20 +285,18 @@ func updatePricing() {
 		modelSupportEndpointsStr[ability.Model] = endpoints
 	}
-	// 再补充模型自定义端点：若配置有效则替换默认端点，不做合并
+	// 再补充模型自定义端点：若配置有效则追加到已有推断，不再裁剪渠道真实能力
 	for modelName, meta := range metaMap {
 		if strings.TrimSpace(meta.Endpoints) == "" {
 			continue
 		}
 		var raw map[string]interface{}
-		if err := json.Unmarshal([]byte(meta.Endpoints), &raw); err == nil {
+		if err := common.Unmarshal([]byte(meta.Endpoints), &raw); err == nil {
-			endpoints := make([]string, 0, len(raw))
+			endpoints := modelSupportEndpointsStr[modelName]
 			for k, v := range raw {
 				switch v.(type) {
 				case string, map[string]interface{}:
-					if !common.StringsContains(endpoints, k) {
+					endpoints = appendPricingEndpoint(endpoints, k)
-						endpoints = append(endpoints, k)
-					}
 				}
 			}
 			if len(endpoints) > 0 {
@@ -264,7 +333,7 @@ func updatePricing() {
 			continue
 		}
 		var raw map[string]interface{}
-		if err := json.Unmarshal([]byte(meta.Endpoints), &raw); err == nil {
+		if err := common.Unmarshal([]byte(meta.Endpoints), &raw); err == nil {
 			for k, v := range raw {
 				switch val := v.(type) {
 				case string:

--- a/model/pricing_endpoint_test.go
+++ b/model/pricing_endpoint_test.go
--- a/output/posters/newapi-40k-stars-light.png
+++ b/output/posters/newapi-40k-stars-light.png
--- a/output/posters/newapi-40k-stars-light.svg
+++ b/output/posters/newapi-40k-stars-light.svg
-<svg width="1600" height="900" viewBox="0 0 1600 900" fill="none" xmlns="http://www.w3.org/2000/svg">
-  <defs>
-    <linearGradient id="bg" x1="140" y1="30" x2="1450" y2="880" gradientUnits="userSpaceOnUse">
-      <stop offset="0" stop-color="#F3FCFF"/>
-      <stop offset="0.46" stop-color="#FFFFFF"/>
-      <stop offset="1" stop-color="#FFF6FD"/>
-    </linearGradient>
-    <linearGradient id="logoGradient" x1="430" y1="252" x2="1110" y2="616" gradientUnits="userSpaceOnUse">
-      <stop offset="0" stop-color="#22D3EE"/>
-      <stop offset="0.48" stop-color="#7C3AED"/>
-      <stop offset="1" stop-color="#F15BB5"/>
-    </linearGradient>
-    <linearGradient id="lineGradient" x1="180" y1="120" x2="1440" y2="780" gradientUnits="userSpaceOnUse">
-      <stop offset="0" stop-color="#22D3EE" stop-opacity="0.28"/>
-      <stop offset="0.52" stop-color="#7C3AED" stop-opacity="0.22"/>
-      <stop offset="1" stop-color="#F15BB5" stop-opacity="0.28"/>
-    </linearGradient>
-    <linearGradient id="softBand" x1="300" y1="190" x2="1320" y2="740" gradientUnits="userSpaceOnUse">
-      <stop offset="0" stop-color="#22D3EE" stop-opacity="0.14"/>
-      <stop offset="0.5" stop-color="#7C3AED" stop-opacity="0.1"/>
-      <stop offset="1" stop-color="#F15BB5" stop-opacity="0.14"/>
-    </linearGradient>
-    <filter id="softShadow" x="-20%" y="-20%" width="140%" height="140%" color-interpolation-filters="sRGB">
-      <feDropShadow dx="0" dy="24" stdDeviation="34" flood-color="#64748B" flood-opacity="0.16"/>
-    </filter>
-    <filter id="logoShadow" x="-30%" y="-30%" width="160%" height="160%" color-interpolation-filters="sRGB">
-      <feDropShadow dx="0" dy="18" stdDeviation="24" flood-color="#7C3AED" flood-opacity="0.18"/>
-    </filter>
-    <clipPath id="logoClip">
-      <circle cx="800" cy="206" r="62"/>
-    </clipPath>
-  </defs>
-  <rect width="1600" height="900" rx="0" fill="url(#bg)"/>
-  <g opacity="0.65">
-    <path d="M-90 654C145 538 294 690 495 594C702 495 678 278 908 252C1129 227 1214 421 1697 277" stroke="url(#lineGradient)" stroke-width="2.4"/>
-    <path d="M-62 266C184 413 331 178 539 280C772 394 782 632 1008 624C1216 617 1332 424 1664 548" stroke="url(#lineGradient)" stroke-width="2"/>
-    <path d="M122 780C392 624 561 779 742 622C905 480 806 323 992 222C1172 124 1322 196 1512 96" stroke="url(#lineGradient)" stroke-width="1.7"/>
-  </g>
-  <g opacity="0.22">
-    <path d="M204 154H1396" stroke="#7C3AED" stroke-width="1"/>
-    <path d="M204 746H1396" stroke="#22D3EE" stroke-width="1"/>
-    <path d="M280 80V820" stroke="#F15BB5" stroke-width="1"/>
-    <path d="M1320 80V820" stroke="#22D3EE" stroke-width="1"/>
-  </g>
-  <g filter="url(#softShadow)">
-    <path d="M310 648C393 543 504 498 637 512C742 524 837 579 941 559C1075 533 1173 414 1291 457C1376 488 1420 580 1428 665C1284 741 1117 784 932 790C702 797 492 750 310 648Z" fill="url(#softBand)"/>
-  </g>
-  <g opacity="0.9">
-    <circle cx="376" cy="232" r="7" fill="#22D3EE"/>
-    <circle cx="1266" cy="236" r="7" fill="#F15BB5"/>
-    <circle cx="1328" cy="642" r="6" fill="#7C3AED"/>
-    <circle cx="254" cy="606" r="5" fill="#F15BB5"/>
-    <circle cx="1172" cy="154" r="4" fill="#22D3EE"/>
-    <circle cx="506" cy="760" r="4" fill="#7C3AED"/>
-  </g>
-  <g opacity="0.82">
-    <path d="M1190 358L1202 386L1232 398L1202 410L1190 438L1178 410L1148 398L1178 386L1190 358Z" fill="#22D3EE"/>
-    <path d="M420 416L430 440L456 450L430 460L420 484L410 460L384 450L410 440L420 416Z" fill="#F15BB5"/>
-    <path d="M1074 718L1083 739L1105 748L1083 757L1074 778L1065 757L1043 748L1065 739L1074 718Z" fill="#7C3AED"/>
-  </g>
-  <g transform="translate(0 10)">
-    <g filter="url(#logoShadow)">
-      <circle cx="800" cy="206" r="78" fill="#FFFFFF"/>
-      <circle cx="800" cy="206" r="77" stroke="#E8F3FF" stroke-width="2"/>
-      <image href="/Users/caion/GolandProjects/new-api/web/default/public/logo.png" x="738" y="144" width="124" height="124" clip-path="url(#logoClip)" preserveAspectRatio="xMidYMid meet"/>
-    </g>
-    <text x="800" y="344" text-anchor="middle"
-      font-family="Inter, ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, Segoe UI, sans-serif"
-      font-size="46" font-weight="760" letter-spacing="0" fill="#0F172A">NewAPI</text>
-    <text x="800" y="516" text-anchor="middle"
-      font-family="Inter, ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, Segoe UI, sans-serif"
-      font-size="170" font-weight="860" letter-spacing="0" fill="url(#logoGradient)">40K</text>
-    <text x="800" y="616" text-anchor="middle"
-      font-family="Inter, ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, Segoe UI, sans-serif"
-      font-size="76" font-weight="780" letter-spacing="0" fill="#101828">Stars</text>
-    <text x="800" y="688" text-anchor="middle"
-      font-family="Inter, ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, Segoe UI, sans-serif"
-      font-size="34" font-weight="560" letter-spacing="0" fill="#475569">Thank you, builders</text>
-  </g>
-  <g transform="translate(104 86)">
-    <image href="/Users/caion/GolandProjects/new-api/web/default/public/logo.png" x="0" y="0" width="42" height="42" preserveAspectRatio="xMidYMid meet"/>
-    <text x="58" y="30"
-      font-family="Inter, ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, Segoe UI, sans-serif"
-      font-size="26" font-weight="730" letter-spacing="0" fill="#162033">NewAPI</text>
-  </g>
-  <g transform="translate(1262 96)">
-    <rect x="0" y="0" width="234" height="54" rx="27" fill="#FFFFFF" fill-opacity="0.78" stroke="#D7E7F5"/>
-    <text x="117" y="35" text-anchor="middle"
-      font-family="Inter, ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, Segoe UI, sans-serif"
-      font-size="23" font-weight="660" letter-spacing="0" fill="#334155">newapi.ai</text>
-  </g>
-  <g transform="translate(118 774)">
-    <path d="M0 22H246" stroke="#22D3EE" stroke-width="5" stroke-linecap="round"/>
-    <path d="M281 22H464" stroke="#7C3AED" stroke-width="5" stroke-linecap="round"/>
-    <path d="M500 22H652" stroke="#F15BB5" stroke-width="5" stroke-linecap="round"/>
-  </g>
-</svg>
--- a/relay/channel/advancedcustom/adaptor.go
+++ b/relay/channel/advancedcustom/adaptor.go
--- a/relay/channel/advancedcustom/adaptor_test.go
+++ b/relay/channel/advancedcustom/adaptor_test.go
--- a/relay/channel/ali/adaptor.go
+++ b/relay/channel/ali/adaptor.go
@@ -75,10 +75,14 @@ func (a *Adaptor) ConvertClaudeRequest(c *gin.Context, info *relaycommon.RelayIn
 		return req, nil
 	}
-	oaiReq, err := service.ClaudeToOpenAIRequest(*req, info)
+	result, err := service.ConvertRequest(c, info, types.RelayFormatOpenAI, req)
 	if err != nil {
 		return nil, err
 	}
+	oaiReq, ok := result.Value.(*dto.GeneralOpenAIRequest)
+	if !ok {
+		return nil, fmt.Errorf("expected OpenAI chat completions request, got %T", result.Value)
+	}
 	if info.SupportStreamOptions && info.IsStream {
 		oaiReq.StreamOptions = &dto.StreamOptions{IncludeUsage: true}
 	}

--- a/relay/channel/api_request.go
+++ b/relay/channel/api_request.go
@@ -309,7 +309,7 @@ func DoApiRequest(a Adaptor, c *gin.Context, info *common.RelayInfo, requestBody
 	if err != nil {
 		return nil, fmt.Errorf("get request url failed: %w", err)
 	}
-	logger.LogDebug(c, "fullRequestURL: %s", fullRequestURL)
+	logger.LogDebug(c, "fullRequestURL: %s", common.SanitizeURLForLog(fullRequestURL))
 	req, err := http.NewRequest(c.Request.Method, fullRequestURL, requestBody)
 	if err != nil {
 		return nil, fmt.Errorf("new request failed: %w", err)
@@ -339,7 +339,7 @@ func DoFormRequest(a Adaptor, c *gin.Context, info *common.RelayInfo, requestBod
 	if err != nil {
 		return nil, fmt.Errorf("get request url failed: %w", err)
 	}
-	logger.LogDebug(c, "fullRequestURL: %s", fullRequestURL)
+	logger.LogDebug(c, "fullRequestURL: %s", common.SanitizeURLForLog(fullRequestURL))
 	req, err := http.NewRequest(c.Request.Method, fullRequestURL, requestBody)
 	if err != nil {
 		return nil, fmt.Errorf("new request failed: %w", err)
@@ -388,7 +388,7 @@ func DoWssRequest(a Adaptor, c *gin.Context, info *common.RelayInfo, requestBody
 	targetHeader.Set("Content-Type", c.Request.Header.Get("Content-Type"))
 	targetConn, _, err := websocket.DefaultDialer.Dial(fullRequestURL, targetHeader)
 	if err != nil {
-		return nil, fmt.Errorf("dial failed to %s: %w", fullRequestURL, err)
+		return nil, fmt.Errorf("dial failed to %s: %w", common.SanitizeURLForLog(fullRequestURL), err)
 	}
 	// send request body
 	//all, err := io.ReadAll(requestBody)

--- a/relay/channel/aws/adaptor.go
+++ b/relay/channel/aws/adaptor.go
@@ -123,10 +123,14 @@ func (a *Adaptor) ConvertOpenAIRequest(c *gin.Context, info *relaycommon.RelayIn
 	}
 	// 原有的Claude模型处理逻辑
-	claudeReq, err := claude.RequestOpenAI2ClaudeMessage(c, *request)
+	result, err := service.ConvertRequest(c, info, types.RelayFormatClaude, request)
 	if err != nil {
 		return nil, errors.Wrap(err, "failed to convert openai request to claude request")
 	}
+	claudeReq, ok := result.Value.(*dto.ClaudeRequest)
+	if !ok {
+		return nil, fmt.Errorf("expected Anthropic Messages request, got %T", result.Value)
+	}
 	info.UpstreamModelName = claudeReq.Model
 	return claudeReq, err
 }

--- a/relay/channel/claude/adaptor.go
+++ b/relay/channel/claude/adaptor.go
@@ -10,6 +10,7 @@ import (
 	"github.com/QuantumNous/new-api/dto"
 	"github.com/QuantumNous/new-api/relay/channel"
 	relaycommon "github.com/QuantumNous/new-api/relay/common"
+	"github.com/QuantumNous/new-api/service/relayconvert"
 	"github.com/QuantumNous/new-api/setting/model_setting"
 	"github.com/QuantumNous/new-api/types"
@@ -95,7 +96,11 @@ func (a *Adaptor) ConvertOpenAIRequest(c *gin.Context, info *relaycommon.RelayIn
 	if request == nil {
 		return nil, errors.New("request is nil")
 	}
-	return RequestOpenAI2ClaudeMessage(c, *request)
+	result, err := relayconvert.ConvertRequest(c, info, types.RelayFormatClaude, request)
+	if err != nil {
+		return nil, err
+	}
+	return result.Value, nil
 }
 func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dto.RerankRequest) (any, error) {

--- a/relay/channel/claude/relay-claude.go
+++ b/relay/channel/claude/relay-claude.go
--- a/relay/channel/claude/relay_claude_test.go
+++ b/relay/channel/claude/relay_claude_test.go
@@ -5,7 +5,7 @@ import (
 	"testing"
 	"github.com/QuantumNous/new-api/dto"
-	"github.com/QuantumNous/new-api/service"
+	"github.com/QuantumNous/new-api/service/relayconvert"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 )
@@ -41,7 +41,7 @@ func TestResponseOpenAI2ClaudeToolUseInputIsObject(t *testing.T) {
 					},
 				},
 			})
-			resp := service.ResponseOpenAI2Claude(&dto.OpenAITextResponse{
+			resp := relayconvert.ResponseOpenAI2Claude(&dto.OpenAITextResponse{
 				Id:    "chatcmpl_1",
 				Model: "gpt-test",
 				Choices: []dto.OpenAITextResponseChoice{
@@ -322,7 +322,7 @@ func TestBuildOpenAIStyleUsageFromClaudeUsageDefaultsAggregateCacheCreationTo5m(
 	require.Equal(t, 0, openAIUsage.ClaudeCacheCreation1hTokens)
 }
-func TestRequestOpenAI2ClaudeMessage_ClaudeOpus48HighUsesAdaptiveThinking(t *testing.T) {
+func TestOpenAIChatRequestToClaudeMessages_ClaudeOpus48HighUsesAdaptiveThinking(t *testing.T) {
 	request := dto.GeneralOpenAIRequest{
 		Model:       "claude-opus-4-8-high",
 		Temperature: commonPointer(0.7),
@@ -336,7 +336,7 @@ func TestRequestOpenAI2ClaudeMessage_ClaudeOpus48HighUsesAdaptiveThinking(t *tes
 		},
 	}
-	claudeRequest, err := RequestOpenAI2ClaudeMessage(nil, request)
+	claudeRequest, err := relayconvert.OpenAIChatRequestToClaudeMessages(nil, request)
 	require.NoError(t, err)
 	require.Equal(t, "claude-opus-4-8", claudeRequest.Model)
 	require.NotNil(t, claudeRequest.Thinking)
@@ -348,7 +348,7 @@ func TestRequestOpenAI2ClaudeMessage_ClaudeOpus48HighUsesAdaptiveThinking(t *tes
 	require.Nil(t, claudeRequest.TopK)
 }
-func TestRequestOpenAI2ClaudeMessage_ClaudeOpus48ThinkingUsesAdaptiveHighEffort(t *testing.T) {
+func TestOpenAIChatRequestToClaudeMessages_ClaudeOpus48ThinkingUsesAdaptiveHighEffort(t *testing.T) {
 	request := dto.GeneralOpenAIRequest{
 		Model:       "claude-opus-4-8-thinking",
 		Temperature: commonPointer(0.7),
@@ -362,7 +362,7 @@ func TestRequestOpenAI2ClaudeMessage_ClaudeOpus48ThinkingUsesAdaptiveHighEffort(
 		},
 	}
-	claudeRequest, err := RequestOpenAI2ClaudeMessage(nil, request)
+	claudeRequest, err := relayconvert.OpenAIChatRequestToClaudeMessages(nil, request)
 	require.NoError(t, err)
 	require.Equal(t, "claude-opus-4-8", claudeRequest.Model)
 	require.NotNil(t, claudeRequest.Thinking)

--- a/relay/channel/gemini/adaptor.go
+++ b/relay/channel/gemini/adaptor.go
@@ -9,7 +9,6 @@ import (
 	"github.com/QuantumNous/new-api/dto"
 	"github.com/QuantumNous/new-api/relay/channel"
-	"github.com/QuantumNous/new-api/relay/channel/openai"
 	relaycommon "github.com/QuantumNous/new-api/relay/common"
 	"github.com/QuantumNous/new-api/relay/constant"
 	"github.com/QuantumNous/new-api/service/relayconvert"
@@ -45,12 +44,15 @@ func (a *Adaptor) ConvertGeminiRequest(c *gin.Context, info *relaycommon.RelayIn
 }
 func (a *Adaptor) ConvertClaudeRequest(c *gin.Context, info *relaycommon.RelayInfo, req *dto.ClaudeRequest) (any, error) {
-	adaptor := openai.Adaptor{}
+	result, err := relayconvert.ConvertRequest(c, info, types.RelayFormatGemini, req)
-	oaiReq, err := adaptor.ConvertClaudeRequest(c, info, req)
 	if err != nil {
 		return nil, err
 	}
-	return a.ConvertOpenAIRequest(c, info, oaiReq.(*dto.GeneralOpenAIRequest))
+	geminiRequest, ok := result.Value.(*dto.GeminiChatRequest)
+	if !ok {
+		return nil, fmt.Errorf("expected Gemini generateContent request, got %T", result.Value)
+	}
+	return geminiRequest, nil
 }
 func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {
@@ -181,13 +183,11 @@ func (a *Adaptor) ConvertOpenAIRequest(c *gin.Context, info *relaycommon.RelayIn
 	if request == nil {
 		return nil, errors.New("request is nil")
 	}
+	result, err := relayconvert.ConvertRequest(c, info, types.RelayFormatGemini, request)
-	geminiRequest, err := CovertOpenAI2Gemini(c, *request, info)
 	if err != nil {
 		return nil, err
 	}
+	return result.Value, nil
-	return geminiRequest, nil
 }
 func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dto.RerankRequest) (any, error) {
@@ -239,17 +239,15 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 }
 func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
-	request, err := preprocessGeminiOpenAIResponsesRequest(request)
+	result, err := relayconvert.ConvertRequest(c, info, types.RelayFormatGemini, &request)
 	if err != nil {
 		return nil, err
 	}
+	geminiRequest, ok := result.Value.(*dto.GeminiChatRequest)
-	chatRequest, err := relayconvert.ResponsesRequestToChatCompletionsRequest(&request)
+	if !ok {
-	if err != nil {
+		return nil, fmt.Errorf("expected Gemini generateContent request, got %T", result.Value)
-		return nil, err
 	}
+	return geminiRequest, nil
-	return a.ConvertOpenAIRequest(c, info, chatRequest)
 }
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {

--- a/relay/channel/gemini/relay-gemini-native.go
+++ b/relay/channel/gemini/relay-gemini-native.go
@@ -39,8 +39,8 @@ func GeminiTextGenerationHandler(c *gin.Context, info *relaycommon.RelayInfo, re
 		common.SetContextKey(c, constant.ContextKeyAdminRejectReason, fmt.Sprintf("gemini_block_reason=%s", *geminiResponse.PromptFeedback.BlockReason))
 	}
-	// 计算使用量（基于 UsageMetadata）
+	// 计算使用量（优先上游 UsageMetadata，缺失时本地估算并保留 Gemini 计费语义）
-	usage := buildUsageFromGeminiMetadata(geminiResponse.UsageMetadata, info.GetEstimatePromptTokens())
+	usage := buildUsageFromGeminiResponse(c, info, &geminiResponse)
 	service.IOCopyBytesGracefully(c, resp, responseBody)

--- a/relay/channel/gemini/relay-gemini.go
+++ b/relay/channel/gemini/relay-gemini.go
--- a/relay/channel/gemini/relay_gemini_usage_test.go
+++ b/relay/channel/gemini/relay_gemini_usage_test.go
@@ -331,3 +331,186 @@ func TestGeminiTextGenerationHandlerUsesEstimatedPromptTokensWhenUsagePromptMiss
 	require.Equal(t, 100, usage.CompletionTokens)
 	require.Equal(t, 110, usage.TotalTokens)
 }
+func TestGeminiChatHandlerMissingUsageMetadataBuildsEstimatedBillingUsage(t *testing.T) {
+	t.Parallel()
+	gin.SetMode(gin.TestMode)
+	c, _ := gin.CreateTestContext(httptest.NewRecorder())
+	c.Request = httptest.NewRequest(http.MethodPost, "/v1/chat/completions", nil)
+	info := &relaycommon.RelayInfo{
+		RelayFormat:     types.RelayFormatGemini,
+		OriginModelName: "gemini-3-flash-preview",
+		ChannelMeta: &relaycommon.ChannelMeta{
+			UpstreamModelName: "gemini-3-flash-preview",
+		},
+	}
+	info.SetEstimatePromptTokens(20)
+	body := []byte(`{"candidates":[{"content":{"role":"model","parts":[{"text":"ok"}]}}]}`)
+	resp := &http.Response{
+		Body: io.NopCloser(bytes.NewReader(body)),
+	}
+	usage, newAPIError := GeminiChatHandler(c, info, resp)
+	require.Nil(t, newAPIError)
+	require.NotNil(t, usage)
+	require.Equal(t, 20, usage.PromptTokens)
+	require.NotNil(t, usage.BillingUsage)
+	require.True(t, usage.BillingUsage.Estimated)
+	require.Equal(t, dto.BillingUsageSourceGeminiChat, usage.BillingUsage.Source)
+	require.Equal(t, dto.BillingUsageSemanticGemini, usage.BillingUsage.Semantic)
+	require.NotNil(t, usage.BillingUsage.GeminiUsageMetadata)
+	require.Equal(t, usage.PromptTokens, usage.BillingUsage.GeminiUsageMetadata.PromptTokenCount)
+	require.Equal(t, usage.CompletionTokens, usage.BillingUsage.GeminiUsageMetadata.CandidatesTokenCount)
+	require.True(t, common.GetContextKeyBool(c, constant.ContextKeyLocalCountTokens))
+}
+func TestGeminiStreamHandlerPromptOnlyUsageMetadataEstimatesCompletionTokens(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	c, _ := gin.CreateTestContext(httptest.NewRecorder())
+	c.Request = httptest.NewRequest(http.MethodPost, "/v1/chat/completions", nil)
+	oldStreamingTimeout := constant.StreamingTimeout
+	constant.StreamingTimeout = 300
+	t.Cleanup(func() {
+		constant.StreamingTimeout = oldStreamingTimeout
+	})
+	info := &relaycommon.RelayInfo{
+		OriginModelName: "gemini-3-flash-preview",
+		ChannelMeta: &relaycommon.ChannelMeta{
+			UpstreamModelName: "gemini-3-flash-preview",
+		},
+	}
+	info.SetEstimatePromptTokens(20)
+	// Simulates a client aborting the stream before the final chunk: text was
+	// streamed but the last observed usageMetadata only carries prompt tokens.
+	chunk := dto.GeminiChatResponse{
+		Candidates: []dto.GeminiChatCandidate{
+			{
+				Content: dto.GeminiChatContent{
+					Role: "model",
+					Parts: []dto.GeminiPart{
+						{Text: "partial streamed answer before disconnect"},
+					},
+				},
+			},
+		},
+		UsageMetadata: dto.GeminiUsageMetadata{
+			PromptTokenCount: 151,
+			TotalTokenCount:  151,
+		},
+	}
+	chunkData, err := common.Marshal(chunk)
+	require.NoError(t, err)
+	streamBody := []byte("data: " + string(chunkData) + "\n" + "data: [DONE]\n")
+	resp := &http.Response{
+		Body: io.NopCloser(bytes.NewReader(streamBody)),
+	}
+	usage, newAPIError := geminiStreamHandler(c, info, resp, func(_ string, _ *dto.GeminiChatResponse) bool {
+		return true
+	})
+	require.Nil(t, newAPIError)
+	require.NotNil(t, usage)
+	require.Equal(t, 151, usage.PromptTokens)
+	require.Greater(t, usage.CompletionTokens, 0)
+	require.Equal(t, usage.PromptTokens+usage.CompletionTokens, usage.TotalTokens)
+	require.NotNil(t, usage.BillingUsage)
+	require.True(t, usage.BillingUsage.Estimated)
+	require.NotNil(t, usage.BillingUsage.GeminiUsageMetadata)
+	require.Equal(t, usage.CompletionTokens, usage.BillingUsage.GeminiUsageMetadata.CandidatesTokenCount)
+}
+func TestGeminiChatHandlerPromptOnlyUsageMetadataEstimatesCompletionTokens(t *testing.T) {
+	t.Parallel()
+	gin.SetMode(gin.TestMode)
+	c, _ := gin.CreateTestContext(httptest.NewRecorder())
+	c.Request = httptest.NewRequest(http.MethodPost, "/v1/chat/completions", nil)
+	info := &relaycommon.RelayInfo{
+		RelayFormat:     types.RelayFormatGemini,
+		OriginModelName: "gemini-3-flash-preview",
+		ChannelMeta: &relaycommon.ChannelMeta{
+			UpstreamModelName: "gemini-3-flash-preview",
+		},
+	}
+	payload := dto.GeminiChatResponse{
+		Candidates: []dto.GeminiChatCandidate{
+			{
+				Content: dto.GeminiChatContent{
+					Role: "model",
+					Parts: []dto.GeminiPart{
+						{Text: "answer text without candidate token count"},
+					},
+				},
+			},
+		},
+		UsageMetadata: dto.GeminiUsageMetadata{
+			PromptTokenCount: 151,
+			TotalTokenCount:  151,
+		},
+	}
+	body, err := common.Marshal(payload)
+	require.NoError(t, err)
+	resp := &http.Response{
+		Body: io.NopCloser(bytes.NewReader(body)),
+	}
+	usage, newAPIError := GeminiChatHandler(c, info, resp)
+	require.Nil(t, newAPIError)
+	require.NotNil(t, usage)
+	require.Equal(t, 151, usage.PromptTokens)
+	require.Greater(t, usage.CompletionTokens, 0)
+	require.Equal(t, usage.PromptTokens+usage.CompletionTokens, usage.TotalTokens)
+	require.NotNil(t, usage.BillingUsage)
+	require.True(t, usage.BillingUsage.Estimated)
+}
+func TestGeminiStreamHandlerEmptyUsageMetadataBuildsEstimatedBillingUsage(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	c, _ := gin.CreateTestContext(httptest.NewRecorder())
+	c.Request = httptest.NewRequest(http.MethodPost, "/v1/chat/completions", nil)
+	oldStreamingTimeout := constant.StreamingTimeout
+	constant.StreamingTimeout = 300
+	t.Cleanup(func() {
+		constant.StreamingTimeout = oldStreamingTimeout
+	})
+	info := &relaycommon.RelayInfo{
+		OriginModelName: "gemini-3-flash-preview",
+		ChannelMeta: &relaycommon.ChannelMeta{
+			UpstreamModelName: "gemini-3-flash-preview",
+		},
+	}
+	info.SetEstimatePromptTokens(20)
+	streamBody := []byte("data: {\"candidates\":[{\"content\":{\"role\":\"model\",\"parts\":[{\"text\":\"partial\"}]}}],\"usageMetadata\":{}}\n" + "data: [DONE]\n")
+	resp := &http.Response{
+		Body: io.NopCloser(bytes.NewReader(streamBody)),
+	}
+	usage, newAPIError := geminiStreamHandler(c, info, resp, func(_ string, _ *dto.GeminiChatResponse) bool {
+		return true
+	})
+	require.Nil(t, newAPIError)
+	require.NotNil(t, usage)
+	require.Equal(t, 20, usage.PromptTokens)
+	require.NotNil(t, usage.BillingUsage)
+	require.True(t, usage.BillingUsage.Estimated)
+	require.Equal(t, dto.BillingUsageSourceGeminiChat, usage.BillingUsage.Source)
+	require.NotNil(t, usage.BillingUsage.GeminiUsageMetadata)
+	require.Equal(t, usage.PromptTokens, usage.BillingUsage.GeminiUsageMetadata.PromptTokenCount)
+	require.Equal(t, usage.CompletionTokens, usage.BillingUsage.GeminiUsageMetadata.CandidatesTokenCount)
+	require.True(t, common.GetContextKeyBool(c, constant.ContextKeyLocalCountTokens))
+}
--- a/relay/channel/gemini/relay_responses.go
+++ b/relay/channel/gemini/relay_responses.go
@@ -32,7 +32,7 @@ func GeminiResponsesHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *h
 		return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
 	}
 	if len(geminiResponse.Candidates) == 0 {
-		usage := buildUsageFromGeminiMetadata(geminiResponse.UsageMetadata, info.GetEstimatePromptTokens())
+		usage := buildUsageFromGeminiResponse(c, info, &geminiResponse)
 		if geminiResponse.PromptFeedback != nil && geminiResponse.PromptFeedback.BlockReason != nil {
 			common.SetContextKey(c, constant.ContextKeyAdminRejectReason, fmt.Sprintf("gemini_block_reason=%s", *geminiResponse.PromptFeedback.BlockReason))
 			return &usage, types.NewOpenAIError(
@@ -51,13 +51,21 @@ func GeminiResponsesHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *h
 	chatResp := responseGeminiChat2OpenAI(c, &geminiResponse)
 	chatResp.Model = info.UpstreamModelName
-	usage := buildUsageFromGeminiMetadata(geminiResponse.UsageMetadata, info.GetEstimatePromptTokens())
+	if responseID := helper.GetResponseID(c); responseID != "" {
+		chatResp.Id = responseID
+	}
+	usage := buildUsageFromGeminiResponse(c, info, &geminiResponse)
 	chatResp.Usage = usage
-	responsesResp, responsesUsage, err := service.ChatCompletionsResponseToResponsesResponse(chatResp, helper.GetResponseID(c))
+	convertResult, err := relayconvert.ConvertResponse(c, info, types.RelayFormatOpenAIResponses, chatResp)
 	if err != nil {
 		return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
 	}
+	responsesResp, ok := convertResult.Value.(*dto.OpenAIResponsesResponse)
+	if !ok {
+		return nil, types.NewOpenAIError(fmt.Errorf("expected OpenAI responses response, got %T", convertResult.Value), types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
+	}
+	responsesUsage := convertResult.Usage
 	if responsesUsage == nil || responsesUsage.TotalTokens == 0 {
 		responsesResp.Usage = relayconvert.UsageFromChatUsage(&usage)
 	}
@@ -73,8 +81,14 @@ func GeminiResponsesHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *h
 func GeminiResponsesStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.Response) (*dto.Usage, *types.NewAPIError) {
 	responseID := helper.GetResponseID(c)
 	created := common.GetTimestamp()
-	state := relayconvert.NewChatToResponsesStreamState(responseID, info.UpstreamModelName)
+	state, err := relayconvert.NewResponseStreamState(types.RelayFormatOpenAI, types.RelayFormatOpenAIResponses, relayconvert.ResponseStreamOptions{
-	state.Created = created
+		ID:      responseID,
+		Model:   info.UpstreamModelName,
+		Created: created,
+	})
+	if err != nil {
+		return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
+	}
 	finishReason := constant.FinishReasonStop
 	toolCallIndexByChoice := make(map[int]map[string]int)
 	nextToolCallIndexByChoice := make(map[int]int)
@@ -90,12 +104,17 @@ func GeminiResponsesStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, r
 		return true
 	}
 	sendChunk := func(chunk *dto.ChatCompletionsStreamResponse) bool {
-		events, err := relayconvert.ChatCompletionsStreamChunkToResponsesEvents(chunk, state)
+		results, err := relayconvert.ConvertStreamResponseChunk(c, info, state, chunk)
 		if err != nil {
 			streamErr = types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
 			return false
 		}
-		for _, event := range events {
+		for _, result := range results {
+			event, ok := result.Value.(relayconvert.ChatToResponsesStreamEvent)
+			if !ok {
+				streamErr = types.NewOpenAIError(fmt.Errorf("expected OAI responses stream event, got %T", result.Value), types.ErrorCodeBadResponse, http.StatusInternalServerError)
+				return false
+			}
 			if !sendEvent(event) {
 				return false
 			}
@@ -103,7 +122,7 @@ func GeminiResponsesStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, r
 		return true
 	}
-	usage, err := geminiStreamHandler(c, info, resp, func(data string, geminiResponse *dto.GeminiChatResponse) bool {
+	usage, streamAPIError := geminiStreamHandler(c, info, resp, func(data string, geminiResponse *dto.GeminiChatResponse) bool {
 		response, isStop := streamResponseGeminiChat2OpenAI(geminiResponse)
 		response.Id = responseID
 		response.Created = created
@@ -143,17 +162,25 @@ func GeminiResponsesStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, r
 		}
 		return true
 	})
-	if err != nil {
+	if streamAPIError != nil {
-		return usage, err
+		return usage, streamAPIError
 	}
 	if streamErr != nil {
 		return nil, streamErr
 	}
 	if usage != nil {
-		state.Usage = relayconvert.UsageFromChatUsage(usage)
+		state.SetUsage(usage)
 	}
-	for _, event := range relayconvert.FinalizeChatCompletionsStreamToResponses(state) {
+	finalResults, err := relayconvert.FinalizeStreamResponse(c, info, state)
+	if err != nil {
+		return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
+	}
+	for _, result := range finalResults {
+		event, ok := result.Value.(relayconvert.ChatToResponsesStreamEvent)
+		if !ok {
+			return nil, types.NewOpenAIError(fmt.Errorf("expected OAI responses stream event, got %T", result.Value), types.ErrorCodeBadResponse, http.StatusInternalServerError)
+		}
 		if !sendEvent(event) {
 			return nil, streamErr
 		}

--- a/relay/channel/openai/adaptor.go
+++ b/relay/channel/openai/adaptor.go
@@ -42,11 +42,14 @@ type Adaptor struct {
 }
 func (a *Adaptor) ConvertGeminiRequest(c *gin.Context, info *relaycommon.RelayInfo, request *dto.GeminiChatRequest) (any, error) {
-	// 使用 service.GeminiToOpenAIRequest 转换请求格式
+	result, err := service.ConvertRequest(c, info, types.RelayFormatOpenAI, request)
-	openaiRequest, err := service.GeminiToOpenAIRequest(request, info)
 	if err != nil {
 		return nil, err
 	}
+	openaiRequest, ok := result.Value.(*dto.GeneralOpenAIRequest)
+	if !ok {
+		return nil, fmt.Errorf("expected OpenAI chat completions request, got %T", result.Value)
+	}
 	return a.ConvertOpenAIRequest(c, info, openaiRequest)
 }
@@ -61,10 +64,14 @@ func (a *Adaptor) ConvertClaudeRequest(c *gin.Context, info *relaycommon.RelayIn
 	//		println(fmt.Sprintf("failed to save request body to file: %v", err))
 	//	}
 	//}
-	aiRequest, err := service.ClaudeToOpenAIRequest(*request, info)
+	result, err := service.ConvertRequest(c, info, types.RelayFormatOpenAI, request)
 	if err != nil {
 		return nil, err
 	}
+	aiRequest, ok := result.Value.(*dto.GeneralOpenAIRequest)
+	if !ok {
+		return nil, fmt.Errorf("expected OpenAI chat completions request, got %T", result.Value)
+	}
 	//if common.DebugEnabled {
 	//	println(fmt.Sprintf("convert claude to openai request result: %s", common.GetJsonString(aiRequest)))
 	//	// Save request body to file for debugging

--- a/relay/channel/openai/chat_via_responses.go
+++ b/relay/channel/openai/chat_via_responses.go
@@ -41,11 +41,18 @@ func OaiResponsesToChatHandler(c *gin.Context, info *relaycommon.RelayInfo, resp
 		return nil, types.WithOpenAIError(*oaiError, resp.StatusCode)
 	}
-	chatId := helper.GetResponseID(c)
+	chatResult, err := relayconvert.ConvertResponse(c, info, types.RelayFormatOpenAI, &responsesResp)
-	chatResp, usage, err := service.ResponsesResponseToChatCompletionsResponse(&responsesResp, chatId)
 	if err != nil {
 		return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
 	}
+	chatResp, ok := chatResult.Value.(*dto.OpenAITextResponse)
+	if !ok {
+		return nil, types.NewOpenAIError(fmt.Errorf("expected OpenAI chat response, got %T", chatResult.Value), types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
+	}
+	if chatID := helper.GetResponseID(c); chatID != "" {
+		chatResp.Id = chatID
+	}
+	usage := chatResult.Usage
 	if usage == nil || usage.TotalTokens == 0 {
 		text := service.ExtractOutputTextFromResponses(&responsesResp)
@@ -53,17 +60,15 @@ func OaiResponsesToChatHandler(c *gin.Context, info *relaycommon.RelayInfo, resp
 		chatResp.Usage = *usage
 	}
-	var responseBody []byte
+	responseValue := any(chatResp)
-	switch info.RelayFormat {
+	if info.RelayFormat != types.RelayFormatOpenAI {
-	case types.RelayFormatClaude:
+		targetResult, err := relayconvert.ConvertResponse(c, info, info.RelayFormat, chatResp)
-		claudeResp := service.ResponseOpenAI2Claude(chatResp, info)
+		if err != nil {
-		responseBody, err = common.Marshal(claudeResp)
+			return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
-	case types.RelayFormatGemini:
+		}
-		geminiResp := service.ResponseOpenAI2Gemini(chatResp, info)
+		responseValue = targetResult.Value
-		responseBody, err = common.Marshal(geminiResp)
-	default:
-		responseBody, err = common.Marshal(chatResp)
 	}
+	responseBody, err := common.Marshal(responseValue)
 	if err != nil {
 		return nil, types.NewOpenAIError(err, types.ErrorCodeJsonMarshalFailed, http.StatusInternalServerError)
 	}
@@ -145,28 +150,33 @@ func OaiResponsesToChatBufferedStreamHandler(c *gin.Context, info *relaycommon.R
 	}
 	accumulator.SupplementResponseOutput(finalResponse)
-	chatId := helper.GetResponseID(c)
+	chatResult, err := relayconvert.ConvertResponse(c, info, types.RelayFormatOpenAI, finalResponse)
-	chatResp, usage, err := service.ResponsesResponseToChatCompletionsResponse(finalResponse, chatId)
 	if err != nil {
 		return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
 	}
+	chatResp, ok := chatResult.Value.(*dto.OpenAITextResponse)
+	if !ok {
+		return nil, types.NewOpenAIError(fmt.Errorf("expected OpenAI chat response, got %T", chatResult.Value), types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
+	}
+	if chatID := helper.GetResponseID(c); chatID != "" {
+		chatResp.Id = chatID
+	}
+	usage := chatResult.Usage
 	if usage == nil || usage.TotalTokens == 0 {
 		text := service.ExtractOutputTextFromResponses(finalResponse)
 		usage = service.ResponseText2Usage(c, text, info.UpstreamModelName, info.GetEstimatePromptTokens())
 		chatResp.Usage = *usage
 	}
-	var responseBody []byte
+	responseValue := any(chatResp)
-	switch info.RelayFormat {
+	if info.RelayFormat != types.RelayFormatOpenAI {
-	case types.RelayFormatClaude:
+		targetResult, err := relayconvert.ConvertResponse(c, info, info.RelayFormat, chatResp)
-		claudeResp := service.ResponseOpenAI2Claude(chatResp, info)
+		if err != nil {
-		responseBody, err = common.Marshal(claudeResp)
+			return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
-	case types.RelayFormatGemini:
+		}
-		geminiResp := service.ResponseOpenAI2Gemini(chatResp, info)
+		responseValue = targetResult.Value
-		responseBody, err = common.Marshal(geminiResp)
-	default:
-		responseBody, err = common.Marshal(chatResp)
 	}
+	responseBody, err := common.Marshal(responseValue)
 	if err != nil {
 		return nil, types.NewOpenAIError(err, types.ErrorCodeJsonMarshalFailed, http.StatusInternalServerError)
 	}
@@ -184,37 +194,77 @@ func OaiResponsesToChatStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 	responseId := helper.GetResponseID(c)
 	createAt := time.Now().Unix()
-	state := relayconvert.NewResponsesToChatStreamState(info.UpstreamModelName, false)
+	state, err := relayconvert.NewResponseStreamState(types.RelayFormatOpenAIResponses, info.RelayFormat, relayconvert.ResponseStreamOptions{
-	state.ID = responseId
+		ID:      responseId,
-	state.Created = createAt
+		Model:   info.UpstreamModelName,
+		Created: createAt,
+	})
+	if err != nil {
+		return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
+	}
 	streamErr := (*types.NewAPIError)(nil)
 	if info.RelayFormat == types.RelayFormatClaude && info.ClaudeConvertInfo == nil {
 		info.ClaudeConvertInfo = &relaycommon.ClaudeConvertInfo{LastMessagesType: relaycommon.LastMessageTypeNone}
 	}
-	sendChatChunk := func(chunk dto.ChatCompletionsStreamResponse) bool {
+	sendGeminiResponse := func(geminiResponse *dto.GeminiChatResponse) bool {
-		if len(chunk.Choices) == 0 && chunk.Usage == nil {
+		if geminiResponse == nil {
 			return true
 		}
-		if info.RelayFormat == types.RelayFormatOpenAI {
+		geminiResponseStr, err := common.Marshal(geminiResponse)
-			if err := helper.ObjectData(c, &chunk); err != nil {
-				streamErr = types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
-				return false
-			}
-			return true
-		}
-		chunkData, err := common.Marshal(&chunk)
 		if err != nil {
 			streamErr = types.NewOpenAIError(err, types.ErrorCodeJsonMarshalFailed, http.StatusInternalServerError)
 			return false
 		}
-		if err := HandleStreamFormat(c, info, string(chunkData), false, false); err != nil {
+		c.Render(-1, common.CustomEvent{Data: "data: " + string(geminiResponseStr)})
-			streamErr = types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
+		_ = helper.FlushWriter(c)
+		return true
+	}
+	sendStreamResult := func(result relayconvert.ResponseResult) bool {
+		switch value := result.Value.(type) {
+		case dto.ChatCompletionsStreamResponse:
+			if len(value.Choices) == 0 && value.Usage == nil {
+				return true
+			}
+			if err := helper.ObjectData(c, &value); err != nil {
+				streamErr = types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
+				return false
+			}
+			return true
+		case *dto.ChatCompletionsStreamResponse:
+			if value == nil || (len(value.Choices) == 0 && value.Usage == nil) {
+				return true
+			}
+			if err := helper.ObjectData(c, value); err != nil {
+				streamErr = types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
+				return false
+			}
+			return true
+		case dto.ClaudeResponse:
+			if err := helper.ClaudeData(c, value); err != nil {
+				streamErr = types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
+				return false
+			}
+			return true
+		case *dto.ClaudeResponse:
+			if value == nil {
+				return true
+			}
+			if err := helper.ClaudeData(c, *value); err != nil {
+				streamErr = types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
+				return false
+			}
+			return true
+		case dto.GeminiChatResponse:
+			return sendGeminiResponse(&value)
+		case *dto.GeminiChatResponse:
+			return sendGeminiResponse(value)
+		default:
+			streamErr = types.NewOpenAIError(fmt.Errorf("unsupported converted stream response type %T", result.Value), types.ErrorCodeBadResponse, http.StatusInternalServerError)
 			return false
 		}
-		return true
 	}
 	helper.StreamScannerHandler(c, resp, info, func(data string, sr *helper.StreamResult) {
@@ -243,14 +293,14 @@ func OaiResponsesToChatStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 			return
 		}
-		chunks, err := relayconvert.ResponsesStreamEventToChatChunks(&streamResp, state)
+		results, err := relayconvert.ConvertStreamResponseChunk(c, info, state, &streamResp)
 		if err != nil {
 			streamErr = types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
 			sr.Stop(streamErr)
 			return
 		}
-		for _, chunk := range chunks {
+		for _, result := range results {
-			if !sendChatChunk(chunk) {
+			if !sendStreamResult(result) {
 				sr.Stop(streamErr)
 				return
 			}
@@ -261,22 +311,26 @@ func OaiResponsesToChatStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 		return nil, streamErr
 	}
-	usage := state.Usage
+	usage := state.Usage()
-	if usage.TotalTokens == 0 {
+	if usage == nil || usage.TotalTokens == 0 {
 		usage = service.ResponseText2Usage(c, state.UsageText(), info.UpstreamModelName, info.GetEstimatePromptTokens())
-		state.Usage = usage
+		state.SetUsage(usage)
 	}
 	if info.RelayFormat == types.RelayFormatClaude && info.ClaudeConvertInfo != nil {
 		info.ClaudeConvertInfo.Usage = usage
 	}
-	for _, chunk := range relayconvert.FinalizeResponsesToChatStream(state) {
+	finalResults, err := relayconvert.FinalizeStreamResponse(c, info, state)
-		if !sendChatChunk(chunk) {
+	if err != nil {
+		return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
+	}
+	for _, result := range finalResults {
+		if !sendStreamResult(result) {
 			return nil, streamErr
 		}
 	}
 	if info.RelayFormat == types.RelayFormatOpenAI && info.ShouldIncludeUsage && usage != nil {
-		if err := helper.ObjectData(c, helper.GenerateFinalUsageResponse(responseId, state.Created, state.Model, *usage)); err != nil {
+		if err := helper.ObjectData(c, helper.GenerateFinalUsageResponse(responseId, createAt, info.UpstreamModelName, *usage)); err != nil {
 			return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
 		}
 	}

--- a/relay/channel/openai/helper.go
+++ b/relay/channel/openai/helper.go
 package openai
 import (
+	"fmt"
 	"strings"
 	"github.com/QuantumNous/new-api/common"
@@ -10,6 +11,7 @@ import (
 	relayconstant "github.com/QuantumNous/new-api/relay/constant"
 	"github.com/QuantumNous/new-api/relay/helper"
 	"github.com/QuantumNous/new-api/service"
+	"github.com/QuantumNous/new-api/service/relayconvert"
 	"github.com/QuantumNous/new-api/types"
 	"github.com/samber/lo"
@@ -41,7 +43,14 @@ func handleClaudeFormat(c *gin.Context, data string, info *relaycommon.RelayInfo
 	if streamResponse.Usage != nil {
 		info.ClaudeConvertInfo.Usage = streamResponse.Usage
 	}
-	claudeResponses := service.StreamResponseOpenAI2Claude(&streamResponse, info)
+	result, err := relayconvert.ConvertStreamResponse(c, info, types.RelayFormatClaude, &streamResponse)
+	if err != nil {
+		return err
+	}
+	claudeResponses, ok := result.Value.([]*dto.ClaudeResponse)
+	if !ok {
+		return fmt.Errorf("expected Claude stream responses, got %T", result.Value)
+	}
 	for _, resp := range claudeResponses {
 		helper.ClaudeData(c, *resp)
 	}
@@ -55,7 +64,14 @@ func handleGeminiFormat(c *gin.Context, data string, info *relaycommon.RelayInfo
 		return err
 	}
-	geminiResponse := service.StreamResponseOpenAI2Gemini(&streamResponse, info)
+	result, err := relayconvert.ConvertStreamResponse(c, info, types.RelayFormatGemini, &streamResponse)
+	if err != nil {
+		return err
+	}
+	geminiResponse, ok := result.Value.(*dto.GeminiChatResponse)
+	if !ok {
+		return fmt.Errorf("expected Gemini stream response, got %T", result.Value)
+	}
 	// 如果返回 nil，表示没有实际内容，跳过发送
 	if geminiResponse == nil {
@@ -165,7 +181,16 @@ func HandleFinalResponse(c *gin.Context, info *relaycommon.RelayInfo, lastStream
 		info.ClaudeConvertInfo.Usage = usage
-		claudeResponses := service.StreamResponseOpenAI2Claude(&streamResponse, info)
+		result, err := relayconvert.ConvertStreamResponse(c, info, types.RelayFormatClaude, &streamResponse)
+		if err != nil {
+			common.SysLog("error converting Claude stream response: " + err.Error())
+			return
+		}
+		claudeResponses, ok := result.Value.([]*dto.ClaudeResponse)
+		if !ok {
+			common.SysLog(fmt.Sprintf("expected Claude stream responses, got %T", result.Value))
+			return
+		}
 		for _, resp := range claudeResponses {
 			_ = helper.ClaudeData(c, *resp)
 		}
@@ -183,7 +208,16 @@ func HandleFinalResponse(c *gin.Context, info *relaycommon.RelayInfo, lastStream
 		// 而包含最后一段文本输出的响应（倒数第二个）的 finishReason 为 null
 		// 暂不知是否有程序会不兼容。
-		geminiResponse := service.StreamResponseOpenAI2Gemini(&streamResponse, info)
+		result, err := relayconvert.ConvertStreamResponse(c, info, types.RelayFormatGemini, &streamResponse)
+		if err != nil {
+			common.SysLog("error converting Gemini stream response: " + err.Error())
+			return
+		}
+		geminiResponse, ok := result.Value.(*dto.GeminiChatResponse)
+		if !ok {
+			common.SysLog(fmt.Sprintf("expected Gemini stream response, got %T", result.Value))
+			return
+		}
 		// openai 流响应开头的空数据
 		if geminiResponse == nil {

--- a/relay/channel/openai/relay-openai.go
+++ b/relay/channel/openai/relay-openai.go
@@ -14,6 +14,7 @@ import (
 	relaycommon "github.com/QuantumNous/new-api/relay/common"
 	"github.com/QuantumNous/new-api/relay/helper"
 	"github.com/QuantumNous/new-api/service"
+	"github.com/QuantumNous/new-api/service/relayconvert"
 	"github.com/QuantumNous/new-api/types"
 	"github.com/gin-gonic/gin"
@@ -271,15 +272,21 @@ func OpenaiHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.Respo
 			break
 		}
 	case types.RelayFormatClaude:
-		claudeResp := service.ResponseOpenAI2Claude(&simpleResponse, info)
+		convertResult, err := relayconvert.ConvertResponse(c, info, types.RelayFormatClaude, &simpleResponse)
-		claudeRespStr, err := common.Marshal(claudeResp)
+		if err != nil {
+			return nil, types.NewError(err, types.ErrorCodeBadResponseBody)
+		}
+		claudeRespStr, err := common.Marshal(convertResult.Value)
 		if err != nil {
 			return nil, types.NewError(err, types.ErrorCodeBadResponseBody)
 		}
 		responseBody = claudeRespStr
 	case types.RelayFormatGemini:
-		geminiResp := service.ResponseOpenAI2Gemini(&simpleResponse, info)
+		convertResult, err := relayconvert.ConvertResponse(c, info, types.RelayFormatGemini, &simpleResponse)
-		geminiRespStr, err := common.Marshal(geminiResp)
+		if err != nil {
+			return nil, types.NewError(err, types.ErrorCodeBadResponseBody)
+		}
+		geminiRespStr, err := common.Marshal(convertResult.Value)
 		if err != nil {
 			return nil, types.NewError(err, types.ErrorCodeBadResponseBody)
 		}

--- a/relay/channel/openai/responses_via_chat.go
+++ b/relay/channel/openai/responses_via_chat.go
@@ -35,11 +35,18 @@ func OaiChatToResponsesHandler(c *gin.Context, info *relaycommon.RelayInfo, resp
 		return nil, types.WithOpenAIError(*oaiError, resp.StatusCode)
 	}
-	responseID := helper.GetResponseID(c)
+	if responseID := helper.GetResponseID(c); responseID != "" {
-	responsesResp, usage, err := service.ChatCompletionsResponseToResponsesResponse(&chatResp, responseID)
+		chatResp.Id = responseID
+	}
+	convertResult, err := relayconvert.ConvertResponse(c, info, types.RelayFormatOpenAIResponses, &chatResp)
 	if err != nil {
 		return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
 	}
+	responsesResp, ok := convertResult.Value.(*dto.OpenAIResponsesResponse)
+	if !ok {
+		return nil, types.NewOpenAIError(fmt.Errorf("expected OpenAI responses response, got %T", convertResult.Value), types.ErrorCodeBadResponseBody, http.StatusInternalServerError)
+	}
+	usage := convertResult.Usage
 	if usage == nil || usage.TotalTokens == 0 {
 		text := service.ExtractOutputTextFromResponses(responsesResp)
 		usage = service.ResponseText2Usage(c, text, info.UpstreamModelName, info.GetEstimatePromptTokens())
@@ -62,7 +69,13 @@ func OaiChatToResponsesStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 	defer service.CloseResponseBodyGracefully(resp)
 	responseID := helper.GetResponseID(c)
-	state := relayconvert.NewChatToResponsesStreamState(responseID, info.UpstreamModelName)
+	state, err := relayconvert.NewResponseStreamState(types.RelayFormatOpenAI, types.RelayFormatOpenAIResponses, relayconvert.ResponseStreamOptions{
+		ID:    responseID,
+		Model: info.UpstreamModelName,
+	})
+	if err != nil {
+		return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
+	}
 	streamErr := (*types.NewAPIError)(nil)
 	sendEvent := func(event relayconvert.ChatToResponsesStreamEvent) bool {
@@ -97,13 +110,19 @@ func OaiChatToResponsesStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 			return
 		}
-		events, err := relayconvert.ChatCompletionsStreamChunkToResponsesEvents(&chunk, state)
+		results, err := relayconvert.ConvertStreamResponseChunk(c, info, state, &chunk)
 		if err != nil {
 			streamErr = types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
 			sr.Stop(streamErr)
 			return
 		}
-		for _, event := range events {
+		for _, result := range results {
+			event, ok := result.Value.(relayconvert.ChatToResponsesStreamEvent)
+			if !ok {
+				streamErr = types.NewOpenAIError(fmt.Errorf("expected OAI responses stream event, got %T", result.Value), types.ErrorCodeBadResponse, http.StatusInternalServerError)
+				sr.Stop(streamErr)
+				return
+			}
 			if !sendEvent(event) {
 				sr.Stop(streamErr)
 				return
@@ -115,13 +134,21 @@ func OaiChatToResponsesStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 		return nil, streamErr
 	}
-	usage := state.Usage
+	usage := state.Usage()
 	if usage == nil || usage.TotalTokens == 0 {
 		usage = service.ResponseText2Usage(c, state.UsageText(), info.UpstreamModelName, info.GetEstimatePromptTokens())
-		state.Usage = relayconvert.UsageFromChatUsage(usage)
+		state.SetUsage(usage)
 	}
-	for _, event := range relayconvert.FinalizeChatCompletionsStreamToResponses(state) {
+	finalResults, err := relayconvert.FinalizeStreamResponse(c, info, state)
+	if err != nil {
+		return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponse, http.StatusInternalServerError)
+	}
+	for _, result := range finalResults {
+		event, ok := result.Value.(relayconvert.ChatToResponsesStreamEvent)
+		if !ok {
+			return nil, types.NewOpenAIError(fmt.Errorf("expected OAI responses stream event, got %T", result.Value), types.ErrorCodeBadResponse, http.StatusInternalServerError)
+		}
 		if !sendEvent(event) {
 			return nil, streamErr
 		}

--- a/relay/channel/vertex/adaptor.go
+++ b/relay/channel/vertex/adaptor.go
 package vertex
 import (
-	"encoding/json"
 	"errors"
 	"fmt"
 	"io"
@@ -16,6 +15,7 @@ import (
 	"github.com/QuantumNous/new-api/relay/channel/openai"
 	relaycommon "github.com/QuantumNous/new-api/relay/common"
 	"github.com/QuantumNous/new-api/relay/constant"
+	"github.com/QuantumNous/new-api/service"
 	"github.com/QuantumNous/new-api/setting/model_setting"
 	"github.com/QuantumNous/new-api/setting/reasoning"
 	"github.com/QuantumNous/new-api/types"
@@ -267,7 +267,7 @@ func (a *Adaptor) ConvertOpenAIRequest(c *gin.Context, info *relaycommon.RelayIn
 		}
 		if len(request.ExtraBody) > 0 {
 			var extra map[string]any
-			if err := json.Unmarshal(request.ExtraBody, &extra); err == nil {
+			if err := common.Unmarshal(request.ExtraBody, &extra); err == nil {
 				if n, ok := extra["n"].(float64); ok && n > 0 {
 					imgReq.N = lo.ToPtr(uint(n))
 				}
@@ -289,19 +289,27 @@ func (a *Adaptor) ConvertOpenAIRequest(c *gin.Context, info *relaycommon.RelayIn
 		return a.ConvertImageRequest(c, info, imgReq)
 	}
 	if a.RequestMode == RequestModeClaude {
-		claudeReq, err := claude.RequestOpenAI2ClaudeMessage(c, *request)
+		result, err := service.ConvertRequest(c, info, types.RelayFormatClaude, request)
 		if err != nil {
 			return nil, err
 		}
+		claudeReq, ok := result.Value.(*dto.ClaudeRequest)
+		if !ok {
+			return nil, fmt.Errorf("expected Anthropic Messages request, got %T", result.Value)
+		}
 		vertexClaudeReq := copyRequest(claudeReq, anthropicVersion)
 		c.Set("request_model", claudeReq.Model)
 		info.UpstreamModelName = claudeReq.Model
 		return vertexClaudeReq, nil
 	} else if a.RequestMode == RequestModeGemini {
-		geminiRequest, err := gemini.CovertOpenAI2Gemini(c, *request, info)
+		result, err := service.ConvertRequest(c, info, types.RelayFormatGemini, request)
 		if err != nil {
 			return nil, err
 		}
+		geminiRequest, ok := result.Value.(*dto.GeminiChatRequest)
+		if !ok {
+			return nil, fmt.Errorf("expected Gemini generateContent request, got %T", result.Value)
+		}
 		c.Set("request_model", request.Model)
 		return geminiRequest, nil
 	} else if a.RequestMode == RequestModeOpenSource {

--- a/relay/chat_completions_via_responses.go
+++ b/relay/chat_completions_via_responses.go
 package relay
 import (
+	"fmt"
 	"io"
 	"net/http"
 	"strings"
@@ -92,11 +93,14 @@ func chatCompletionsViaResponses(c *gin.Context, info *relaycommon.RelayInfo, ad
 		return nil, types.NewError(err, types.ErrorCodeChannelParamOverrideInvalid, types.ErrOptionWithSkipRetry())
 	}
-	responsesReq, err := service.ChatCompletionsRequestToResponsesRequest(&overriddenChatReq)
+	result, err := service.ConvertRequestVia(c, info, &overriddenChatReq, types.RelayFormatOpenAI, types.RelayFormatOpenAIResponses)
 	if err != nil {
 		return nil, types.NewErrorWithStatusCode(err, types.ErrorCodeInvalidRequest, http.StatusBadRequest, types.ErrOptionWithSkipRetry())
 	}
-	info.AppendRequestConversion(types.RelayFormatOpenAIResponses)
+	responsesReq, ok := result.Value.(*dto.OpenAIResponsesRequest)
+	if !ok {
+		return nil, types.NewError(fmt.Errorf("expected OpenAI responses request, got %T", result.Value), types.ErrorCodeConvertRequestFailed, types.ErrOptionWithSkipRetry())
+	}
 	savedRelayMode := info.RelayMode
 	savedRequestURLPath := info.RequestURLPath

--- a/relay/claude_handler.go
+++ b/relay/claude_handler.go
@@ -135,10 +135,14 @@ func ClaudeHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *typ
 	if !model_setting.GetGlobalSettings().PassThroughRequestEnabled &&
 		!info.ChannelSetting.PassThroughBodyEnabled &&
 		service.ShouldChatCompletionsUseResponsesGlobal(info.ChannelId, info.ChannelType, info.OriginModelName) {
-		openAIRequest, convErr := service.ClaudeToOpenAIRequest(*request, info)
+		result, convErr := service.ConvertRequest(c, info, types.RelayFormatOpenAI, request)
 		if convErr != nil {
 			return types.NewError(convErr, types.ErrorCodeConvertRequestFailed, types.ErrOptionWithSkipRetry())
 		}
+		openAIRequest, ok := result.Value.(*dto.GeneralOpenAIRequest)
+		if !ok {
+			return types.NewError(fmt.Errorf("expected OpenAI chat completions request, got %T", result.Value), types.ErrorCodeConvertRequestFailed, types.ErrOptionWithSkipRetry())
+		}
 		usage, newApiErr := chatCompletionsViaResponses(c, info, adaptor, openAIRequest)
 		if newApiErr != nil {

--- a/relay/common/relay_utils.go
+++ b/relay/common/relay_utils.go
@@ -3,6 +3,7 @@ package common
 import (
 	"fmt"
 	"net/http"
+	"net/url"
 	"strconv"
 	"strings"
@@ -36,6 +37,67 @@ func GetFullRequestURL(baseURL string, requestURL string, channelType int) strin
 	return fullRequestURL
 }
+func SanitizeURLForLog(rawURL string) string {
+	if rawURL == "" {
+		return rawURL
+	}
+	parsedURL, err := url.Parse(rawURL)
+	if err != nil {
+		return rawURL
+	}
+	query := parsedURL.Query()
+	if len(query) == 0 {
+		return rawURL
+	}
+	changed := false
+	for key := range query {
+		if isSensitiveURLQueryKey(key) {
+			query.Set(key, "***masked***")
+			changed = true
+		}
+	}
+	if !changed {
+		return rawURL
+	}
+	parsedURL.RawQuery = query.Encode()
+	return parsedURL.String()
+}
+func isSensitiveURLQueryKey(key string) bool {
+	normalized := strings.ToLower(strings.TrimSpace(key))
+	switch normalized {
+	case "key",
+		"api_key",
+		"api-key",
+		"apikey",
+		"x-api-key",
+		"access_token",
+		"refresh_token",
+		"id_token",
+		"token",
+		"authorization",
+		"auth",
+		"client_secret",
+		"secret",
+		"password",
+		"passwd",
+		"signature",
+		"sig",
+		"awsaccesskeyid",
+		"x-amz-credential",
+		"x-amz-security-token",
+		"x-amz-signature":
+		return true
+	}
+	return strings.Contains(normalized, "token") ||
+		strings.Contains(normalized, "secret") ||
+		strings.Contains(normalized, "signature")
+}
 func GetAPIVersion(c *gin.Context) string {
 	query := c.Request.URL.Query()
 	apiVersion := query.Get("api-version")

--- a/relay/common/relay_utils_test.go
+++ b/relay/common/relay_utils_test.go
@@ -3,14 +3,59 @@ package common
 import (
 	"net/http"
 	"net/http/httptest"
+	"net/url"
 	"strings"
 	"testing"
 	"github.com/QuantumNous/new-api/constant"
 	"github.com/gin-gonic/gin"
+	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 )
+func TestSanitizeURLForLogMasksSensitiveQueryValues(t *testing.T) {
+	rawURL := "https://example.test/v1beta/models/gemini:streamGenerateContent?alt=sse&key=sk-secret&access_token=ya29-secret&api-version=2024-02-01"
+	got := SanitizeURLForLog(rawURL)
+	assert.NotContains(t, got, "sk-secret")
+	assert.NotContains(t, got, "ya29-secret")
+	parsedURL, err := url.Parse(got)
+	require.NoError(t, err)
+	query := parsedURL.Query()
+	assert.Equal(t, "***masked***", query.Get("key"))
+	assert.Equal(t, "***masked***", query.Get("access_token"))
+	assert.Equal(t, "sse", query.Get("alt"))
+	assert.Equal(t, "2024-02-01", query.Get("api-version"))
+}
+func TestSanitizeURLForLogMasksAWSAndSecretLikeQueryKeys(t *testing.T) {
+	rawURL := "https://example.test/path?X-Amz-Credential=credential&X-Amz-Signature=signature&session_token=session&client_secret=secret&model=gpt-test"
+	got := SanitizeURLForLog(rawURL)
+	assert.NotContains(t, got, "X-Amz-Credential=credential")
+	assert.NotContains(t, got, "X-Amz-Signature=signature")
+	assert.NotContains(t, got, "session_token=session")
+	assert.NotContains(t, got, "client_secret=secret")
+	parsedURL, err := url.Parse(got)
+	require.NoError(t, err)
+	query := parsedURL.Query()
+	assert.Equal(t, "***masked***", query.Get("X-Amz-Credential"))
+	assert.Equal(t, "***masked***", query.Get("X-Amz-Signature"))
+	assert.Equal(t, "***masked***", query.Get("session_token"))
+	assert.Equal(t, "***masked***", query.Get("client_secret"))
+	assert.Equal(t, "gpt-test", query.Get("model"))
+}
+func TestSanitizeURLForLogKeepsURLWithoutSensitiveQuery(t *testing.T) {
+	rawURL := "https://example.test/v1/chat/completions?api-version=2024-02-01&alt=sse"
+	got := SanitizeURLForLog(rawURL)
+	assert.Equal(t, rawURL, got)
+}
 func TestValidateMultipartDirectNormalizesImageField(t *testing.T) {
 	gin.SetMode(gin.TestMode)
 	body := strings.NewReader(`{"model":"wan2.7-i2v","prompt":"animate","image":" https://example.com/first.png "}`)

--- a/relay/gemini_handler.go
+++ b/relay/gemini_handler.go
@@ -10,10 +10,10 @@ import (
 	"github.com/QuantumNous/new-api/constant"
 	"github.com/QuantumNous/new-api/dto"
 	"github.com/QuantumNous/new-api/logger"
-	"github.com/QuantumNous/new-api/relay/channel/gemini"
 	relaycommon "github.com/QuantumNous/new-api/relay/common"
 	"github.com/QuantumNous/new-api/relay/helper"
 	"github.com/QuantumNous/new-api/service"
+	"github.com/QuantumNous/new-api/service/relayconvert"
 	"github.com/QuantumNous/new-api/setting/model_setting"
 	"github.com/QuantumNous/new-api/types"
@@ -84,7 +84,7 @@ func GeminiHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *typ
 			}
 		}
 		if request.GenerationConfig.ThinkingConfig == nil {
-			gemini.ThinkingAdaptor(request, info)
+			relayconvert.ApplyGeminiThinkingConfig(request, info)
 		}
 	}

--- a/service/billing_usage.go
+++ b/service/billing_usage.go
+package service
+import (
+	"strings"
+	"github.com/QuantumNous/new-api/dto"
+)
+const (
+	usageBillingPathLocal              = "local"
+	usageBillingPathUpstream           = "upstream"
+	usageBillingPathOpenAI             = "billing-usage-openai"
+	usageBillingPathOpenAIEstimated    = "billing-usage-openai-estimated"
+	usageBillingPathAnthropic          = "billing-usage-anthropic"
+	usageBillingPathAnthropicEstimated = "billing-usage-anthropic-estimated"
+	usageBillingPathGemini             = "billing-usage-gemini"
+	usageBillingPathGeminiEstimated    = "billing-usage-gemini-estimated"
+)
+func effectiveBillingUsage(usage *dto.Usage) *dto.Usage {
+	if billingUsage, ok := usageFromBillingUsage(usage); ok {
+		return billingUsage
+	}
+	return usage
+}
+func usageBillingPathForLog(isLocalCountTokens bool, usage *dto.Usage) string {
+	if isLocalCountTokens {
+		return usageBillingPathLocal
+	}
+	if usage == nil || usage.BillingUsage == nil {
+		return usageBillingPathUpstream
+	}
+	source := strings.TrimSpace(usage.BillingUsage.Source)
+	semantic := strings.TrimSpace(usage.BillingUsage.Semantic)
+	if strings.EqualFold(source, dto.BillingUsageSourceOAIChat) ||
+		strings.EqualFold(source, dto.BillingUsageSourceOAIResponses) ||
+		strings.EqualFold(semantic, dto.BillingUsageSemanticOpenAI) {
+		if usage.BillingUsage.Estimated {
+			return usageBillingPathOpenAIEstimated
+		}
+		return usageBillingPathOpenAI
+	}
+	if strings.EqualFold(source, dto.BillingUsageSourceClaudeMessages) ||
+		strings.EqualFold(semantic, dto.BillingUsageSemanticAnthropic) {
+		if usage.BillingUsage.Estimated {
+			return usageBillingPathAnthropicEstimated
+		}
+		return usageBillingPathAnthropic
+	}
+	if strings.EqualFold(source, dto.BillingUsageSourceGeminiChat) ||
+		strings.EqualFold(semantic, dto.BillingUsageSemanticGemini) {
+		if usage.BillingUsage.Estimated {
+			return usageBillingPathGeminiEstimated
+		}
+		return usageBillingPathGemini
+	}
+	return usageBillingPathUpstream
+}
+func appendUsageBillingPathForLog(other map[string]interface{}, isLocalCountTokens bool, usage *dto.Usage) {
+	if other == nil {
+		return
+	}
+	adminInfo, ok := other["admin_info"].(map[string]interface{})
+	if !ok || adminInfo == nil {
+		adminInfo = make(map[string]interface{})
+		other["admin_info"] = adminInfo
+	}
+	adminInfo["usage_billing_path"] = usageBillingPathForLog(isLocalCountTokens, usage)
+}
+func usageFromBillingUsage(usage *dto.Usage) (*dto.Usage, bool) {
+	if usage == nil || usage.BillingUsage == nil {
+		return nil, false
+	}
+	billingUsage := usage.BillingUsage
+	source := strings.TrimSpace(billingUsage.Source)
+	semantic := strings.TrimSpace(billingUsage.Semantic)
+	if billingUsage.OpenAIUsage != nil &&
+		(strings.EqualFold(source, dto.BillingUsageSourceOAIChat) ||
+			strings.EqualFold(source, dto.BillingUsageSourceOAIResponses) ||
+			strings.EqualFold(semantic, dto.BillingUsageSemanticOpenAI)) {
+		return usageFromOpenAIBillingUsage(billingUsage), true
+	}
+	if billingUsage.ClaudeUsage != nil &&
+		(strings.EqualFold(source, dto.BillingUsageSourceClaudeMessages) ||
+			strings.EqualFold(semantic, dto.BillingUsageSemanticAnthropic)) {
+		return usageFromClaudeBillingUsage(billingUsage), true
+	}
+	if billingUsage.GeminiUsageMetadata != nil &&
+		(strings.EqualFold(source, dto.BillingUsageSourceGeminiChat) ||
+			strings.EqualFold(semantic, dto.BillingUsageSemanticGemini)) {
+		return usageFromGeminiBillingUsage(billingUsage), true
+	}
+	return nil, false
+}
+func usageFromOpenAIBillingUsage(billingUsage *dto.BillingUsage) *dto.Usage {
+	usage := *billingUsage.OpenAIUsage
+	if usage.PromptTokens == 0 && usage.InputTokens > 0 {
+		usage.PromptTokens = usage.InputTokens
+	}
+	if usage.CompletionTokens == 0 && usage.OutputTokens > 0 {
+		usage.CompletionTokens = usage.OutputTokens
+	}
+	if usage.InputTokens == 0 && usage.PromptTokens > 0 {
+		usage.InputTokens = usage.PromptTokens
+	}
+	if usage.OutputTokens == 0 && usage.CompletionTokens > 0 {
+		usage.OutputTokens = usage.CompletionTokens
+	}
+	if usage.TotalTokens == 0 {
+		usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
+	}
+	usage.UsageSemantic = dto.BillingUsageSemanticOpenAI
+	usage.UsageSource = billingUsage.Source
+	usage.BillingUsage = dto.CloneBillingUsage(billingUsage)
+	return &usage
+}
+func usageFromClaudeBillingUsage(billingUsage *dto.BillingUsage) *dto.Usage {
+	claudeUsage := billingUsage.ClaudeUsage
+	cacheCreation5m := claudeUsage.GetCacheCreation5mTokens()
+	if cacheCreation5m == 0 {
+		cacheCreation5m = claudeUsage.ClaudeCacheCreation5mTokens
+	}
+	cacheCreation1h := claudeUsage.GetCacheCreation1hTokens()
+	if cacheCreation1h == 0 {
+		cacheCreation1h = claudeUsage.ClaudeCacheCreation1hTokens
+	}
+	usage := &dto.Usage{
+		PromptTokens:                claudeUsage.InputTokens,
+		CompletionTokens:            claudeUsage.OutputTokens,
+		TotalTokens:                 claudeUsage.InputTokens + claudeUsage.OutputTokens,
+		InputTokens:                 claudeUsage.InputTokens + claudeUsage.CacheReadInputTokens + claudeUsage.CacheCreationInputTokens,
+		OutputTokens:                claudeUsage.OutputTokens,
+		UsageSemantic:               dto.BillingUsageSemanticAnthropic,
+		UsageSource:                 dto.BillingUsageSourceClaudeMessages,
+		BillingUsage:                dto.CloneBillingUsage(billingUsage),
+		ClaudeCacheCreation5mTokens: cacheCreation5m,
+		ClaudeCacheCreation1hTokens: cacheCreation1h,
+	}
+	usage.PromptTokensDetails.CachedTokens = claudeUsage.CacheReadInputTokens
+	usage.PromptTokensDetails.CachedCreationTokens = claudeUsage.CacheCreationInputTokens
+	return usage
+}
+func usageFromGeminiBillingUsage(billingUsage *dto.BillingUsage) *dto.Usage {
+	metadata := *billingUsage.GeminiUsageMetadata
+	promptTokens := metadata.PromptTokenCount + metadata.ToolUsePromptTokenCount
+	usage := &dto.Usage{
+		PromptTokens:     promptTokens,
+		CompletionTokens: metadata.CandidatesTokenCount + metadata.ThoughtsTokenCount,
+		TotalTokens:      metadata.TotalTokenCount,
+		UsageSemantic:    dto.BillingUsageSemanticGemini,
+		UsageSource:      dto.BillingUsageSourceGeminiChat,
+		BillingUsage:     dto.CloneBillingUsage(billingUsage),
+	}
+	usage.CompletionTokenDetails.ReasoningTokens = metadata.ThoughtsTokenCount
+	usage.PromptTokensDetails.CachedTokens = metadata.CachedContentTokenCount
+	for _, detail := range metadata.PromptTokensDetails {
+		addGeminiInputTokenDetail(&usage.PromptTokensDetails, detail)
+	}
+	for _, detail := range metadata.ToolUsePromptTokensDetails {
+		addGeminiInputTokenDetail(&usage.PromptTokensDetails, detail)
+	}
+	for _, detail := range metadata.CandidatesTokensDetails {
+		switch detail.Modality {
+		case "IMAGE":
+			usage.CompletionTokenDetails.ImageTokens += detail.TokenCount
+		case "AUDIO":
+			usage.CompletionTokenDetails.AudioTokens += detail.TokenCount
+		case "TEXT":
+			usage.CompletionTokenDetails.TextTokens += detail.TokenCount
+		}
+	}
+	if usage.TotalTokens == 0 {
+		usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
+	} else if usage.CompletionTokens <= 0 {
+		usage.CompletionTokens = usage.TotalTokens - usage.PromptTokens
+	}
+	if usage.PromptTokens > 0 && usage.PromptTokensDetails.TextTokens == 0 && usage.PromptTokensDetails.AudioTokens == 0 {
+		usage.PromptTokensDetails.TextTokens = usage.PromptTokens
+	}
+	return usage
+}
+func addGeminiInputTokenDetail(details *dto.InputTokenDetails, detail dto.GeminiPromptTokensDetails) {
+	switch detail.Modality {
+	case "AUDIO":
+		details.AudioTokens += detail.TokenCount
+	case "IMAGE":
+		details.ImageTokens += detail.TokenCount
+	case "TEXT":
+		details.TextTokens += detail.TokenCount
+	}
+}
--- a/service/convert.go
+++ b/service/convert.go
--- a/service/convert_test.go
+++ b/service/convert_test.go
+package service
+import (
+	"testing"
+	"github.com/QuantumNous/new-api/dto"
+	relaycommon "github.com/QuantumNous/new-api/relay/common"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+func TestResponseConverterFacades(t *testing.T) {
+	cache5m, cache1h := NormalizeCacheCreationSplit(10, 3, 2)
+	assert.Equal(t, 8, cache5m)
+	assert.Equal(t, 2, cache1h)
+	chatResp := &dto.OpenAITextResponse{
+		Id:    "chatcmpl_1",
+		Model: "gpt-test",
+		Choices: []dto.OpenAITextResponseChoice{
+			{
+				Message: dto.Message{
+					Role:    "assistant",
+					Content: "hello",
+				},
+				FinishReason: "stop",
+			},
+		},
+	}
+	claudeResp := ResponseOpenAI2Claude(chatResp, &relaycommon.RelayInfo{})
+	require.NotNil(t, claudeResp)
+	assert.Equal(t, "message", claudeResp.Type)
+	geminiResp := ResponseOpenAI2Gemini(chatResp, &relaycommon.RelayInfo{})
+	require.NotNil(t, geminiResp)
+	require.Len(t, geminiResp.Candidates, 1)
+}
+func TestStreamResponseConverterFacades(t *testing.T) {
+	info := &relaycommon.RelayInfo{
+		SendResponseCount: 1,
+		ClaudeConvertInfo: &relaycommon.ClaudeConvertInfo{
+			LastMessagesType: relaycommon.LastMessageTypeNone,
+		},
+	}
+	streamResp := &dto.ChatCompletionsStreamResponse{
+		Id:    "chatcmpl_1",
+		Model: "gpt-test",
+		Choices: []dto.ChatCompletionsStreamResponseChoice{
+			{
+				Delta: dto.ChatCompletionsStreamResponseChoiceDelta{
+					Content: ptrValue("hello"),
+				},
+			},
+		},
+	}
+	claudeResponses := StreamResponseOpenAI2Claude(streamResp, info)
+	require.NotEmpty(t, claudeResponses)
+	geminiResp := StreamResponseOpenAI2Gemini(streamResp, &relaycommon.RelayInfo{})
+	require.NotNil(t, geminiResp)
+	require.Len(t, geminiResp.Candidates, 1)
+}
+func ptrValue[T any](value T) *T {
+	return &value
+}
--- a/service/relayconvert/internal/claude_messages/to_oai_chat_req.go
+++ b/service/relayconvert/internal/claude_messages/to_oai_chat_req.go
+package claudemessages
+import (
+	"fmt"
+	"strings"
+	"github.com/QuantumNous/new-api/common"
+	"github.com/QuantumNous/new-api/constant"
+	"github.com/QuantumNous/new-api/dto"
+	relaycommon "github.com/QuantumNous/new-api/relay/common"
+	relaymeta "github.com/QuantumNous/new-api/service/relayconvert/internal/meta"
+)
+const (
+	webSearchMaxUsesLow    = 1
+	webSearchMaxUsesMedium = 5
+	webSearchMaxUsesHigh   = 10
+)
+type openRouterRequestReasoning struct {
+	Enabled   bool   `json:"enabled"`
+	Effort    string `json:"effort,omitempty"`
+	MaxTokens int    `json:"max_tokens,omitempty"`
+	Exclude   bool   `json:"exclude,omitempty"`
+}
+func ClaudeMessagesRequestToOpenAIChat(claudeRequest dto.ClaudeRequest, info *relaycommon.RelayInfo) (*dto.GeneralOpenAIRequest, error) {
+	openAIRequest := dto.GeneralOpenAIRequest{
+		Model:       claudeRequest.Model,
+		Temperature: claudeRequest.Temperature,
+	}
+	if claudeRequest.MaxTokens != nil {
+		openAIRequest.MaxTokens = common.GetPointer(*claudeRequest.MaxTokens)
+	}
+	if claudeRequest.TopP != nil {
+		openAIRequest.TopP = common.GetPointer(*claudeRequest.TopP)
+	}
+	if claudeRequest.TopK != nil {
+		openAIRequest.TopK = common.GetPointer(*claudeRequest.TopK)
+	}
+	if claudeRequest.Stream != nil {
+		openAIRequest.Stream = common.GetPointer(*claudeRequest.Stream)
+	}
+	isOpenRouter := relaymeta.RelayInfoChannelType(info) == constant.ChannelTypeOpenRouter
+	if isOpenRouter {
+		if effort := claudeRequest.GetEfforts(); effort != "" {
+			effortBytes, _ := common.Marshal(effort)
+			openAIRequest.Verbosity = effortBytes
+		}
+		if claudeRequest.Thinking != nil {
+			var reasoningConfig openRouterRequestReasoning
+			if claudeRequest.Thinking.Type == "enabled" {
+				reasoningConfig = openRouterRequestReasoning{
+					Enabled:   true,
+					MaxTokens: claudeRequest.Thinking.GetBudgetTokens(),
+				}
+			} else if claudeRequest.Thinking.Type == "adaptive" {
+				reasoningConfig = openRouterRequestReasoning{
+					Enabled: true,
+				}
+			}
+			reasoningJSON, err := common.Marshal(reasoningConfig)
+			if err != nil {
+				return nil, fmt.Errorf("failed to marshal reasoning: %w", err)
+			}
+			openAIRequest.Reasoning = reasoningJSON
+		}
+	} else if info != nil {
+		thinkingSuffix := "-thinking"
+		if strings.HasSuffix(info.OriginModelName, thinkingSuffix) &&
+			!strings.HasSuffix(openAIRequest.Model, thinkingSuffix) {
+			openAIRequest.Model = openAIRequest.Model + thinkingSuffix
+		}
+	}
+	if len(claudeRequest.StopSequences) == 1 {
+		openAIRequest.Stop = claudeRequest.StopSequences[0]
+	} else if len(claudeRequest.StopSequences) > 1 {
+		openAIRequest.Stop = claudeRequest.StopSequences
+	}
+	tools, _ := common.Any2Type[[]dto.Tool](claudeRequest.Tools)
+	openAITools := make([]dto.ToolCallRequest, 0)
+	for _, claudeTool := range tools {
+		openAITool := dto.ToolCallRequest{
+			Type: "function",
+			Function: dto.FunctionRequest{
+				Name:        claudeTool.Name,
+				Description: claudeTool.Description,
+				Parameters:  claudeTool.InputSchema,
+			},
+		}
+		openAITools = append(openAITools, openAITool)
+	}
+	openAIRequest.Tools = openAITools
+	openAIMessages := make([]dto.Message, 0)
+	if claudeRequest.System != nil {
+		if claudeRequest.IsStringSystem() && claudeRequest.GetStringSystem() != "" {
+			openAIMessage := dto.Message{
+				Role: "system",
+			}
+			openAIMessage.SetStringContent(claudeRequest.GetStringSystem())
+			openAIMessages = append(openAIMessages, openAIMessage)
+		} else {
+			systems := claudeRequest.ParseSystem()
+			if len(systems) > 0 {
+				openAIMessage := dto.Message{
+					Role: "system",
+				}
+				isOpenRouterClaude := isOpenRouter && strings.HasPrefix(relaymeta.RelayInfoUpstreamModelName(info), "anthropic/claude")
+				if isOpenRouterClaude {
+					systemMediaMessages := make([]dto.MediaContent, 0, len(systems))
+					for _, system := range systems {
+						message := dto.MediaContent{
+							Type:         "text",
+							Text:         system.GetText(),
+							CacheControl: system.CacheControl,
+						}
+						systemMediaMessages = append(systemMediaMessages, message)
+					}
+					openAIMessage.SetMediaContent(systemMediaMessages)
+				} else {
+					systemStr := ""
+					for _, system := range systems {
+						if system.Text != nil {
+							systemStr += *system.Text
+						}
+					}
+					openAIMessage.SetStringContent(systemStr)
+				}
+				openAIMessages = append(openAIMessages, openAIMessage)
+			}
+		}
+	}
+	for _, claudeMessage := range claudeRequest.Messages {
+		openAIMessage := dto.Message{
+			Role: claudeMessage.Role,
+		}
+		if claudeMessage.IsStringContent() {
+			openAIMessage.SetStringContent(claudeMessage.GetStringContent())
+		} else {
+			content, err := claudeMessage.ParseContent()
+			if err != nil {
+				return nil, err
+			}
+			var toolCalls []dto.ToolCallRequest
+			mediaMessages := make([]dto.MediaContent, 0, len(content))
+			for _, mediaMsg := range content {
+				switch mediaMsg.Type {
+				case "text", "input_text":
+					message := dto.MediaContent{
+						Type:         "text",
+						Text:         mediaMsg.GetText(),
+						CacheControl: mediaMsg.CacheControl,
+					}
+					mediaMessages = append(mediaMessages, message)
+				case "image":
+					imageData := fmt.Sprintf("data:%s;base64,%s", mediaMsg.Source.MediaType, mediaMsg.Source.Data)
+					mediaMessage := dto.MediaContent{
+						Type:     "image_url",
+						ImageUrl: &dto.MessageImageUrl{Url: imageData},
+					}
+					mediaMessages = append(mediaMessages, mediaMessage)
+				case "tool_use":
+					toolCall := dto.ToolCallRequest{
+						ID:   mediaMsg.Id,
+						Type: "function",
+						Function: dto.FunctionRequest{
+							Name:      mediaMsg.Name,
+							Arguments: requestToJSONString(mediaMsg.Input),
+						},
+					}
+					toolCalls = append(toolCalls, toolCall)
+				case "tool_result":
+					toolName := mediaMsg.Name
+					if toolName == "" {
+						toolName = claudeRequest.SearchToolNameByToolCallId(mediaMsg.ToolUseId)
+					}
+					oaiToolMessage := dto.Message{
+						Role:       "tool",
+						Name:       &toolName,
+						ToolCallId: mediaMsg.ToolUseId,
+					}
+					if mediaMsg.IsStringContent() {
+						oaiToolMessage.SetStringContent(mediaMsg.GetStringContent())
+					} else {
+						mediaContents := mediaMsg.ParseMediaContent()
+						encodedJSON, _ := common.Marshal(mediaContents)
+						oaiToolMessage.SetStringContent(string(encodedJSON))
+					}
+					openAIMessages = append(openAIMessages, oaiToolMessage)
+				}
+			}
+			if len(toolCalls) > 0 {
+				openAIMessage.SetToolCalls(toolCalls)
+			}
+			if len(mediaMessages) > 0 && len(toolCalls) == 0 {
+				openAIMessage.SetMediaContent(mediaMessages)
+			}
+		}
+		if len(openAIMessage.ParseContent()) > 0 || len(openAIMessage.ToolCalls) > 0 {
+			openAIMessages = append(openAIMessages, openAIMessage)
+		}
+	}
+	openAIRequest.Messages = openAIMessages
+	return &openAIRequest, nil
+}
+func requestToJSONString(v interface{}) string {
+	b, err := common.Marshal(v)
+	if err != nil {
+		return "{}"
+	}
+	return string(b)
+}
--- a/service/relayconvert/internal/claude_messages/to_oai_chat_resp.go
+++ b/service/relayconvert/internal/claude_messages/to_oai_chat_resp.go
--- a/service/relayconvert/internal/gemini_chat/to_oai_chat_req.go
+++ b/service/relayconvert/internal/gemini_chat/to_oai_chat_req.go
+package geminichat
+import (
+	"fmt"
+	"strings"
+	"github.com/QuantumNous/new-api/common"
+	"github.com/QuantumNous/new-api/dto"
+	relaycommon "github.com/QuantumNous/new-api/relay/common"
+	"github.com/QuantumNous/new-api/service/relayconvert/internal/jsonutil"
+	relaymeta "github.com/QuantumNous/new-api/service/relayconvert/internal/meta"
+)
+func GeminiGenerateContentRequestToOpenAIChat(geminiRequest *dto.GeminiChatRequest, info *relaycommon.RelayInfo) (*dto.GeneralOpenAIRequest, error) {
+	modelName := ""
+	isStream := false
+	if info != nil {
+		isStream = info.IsStream
+	}
+	modelName = relaymeta.RelayInfoUpstreamModelName(info)
+	openaiRequest := &dto.GeneralOpenAIRequest{
+		Model:  modelName,
+		Stream: common.GetPointer(isStream),
+	}
+	var messages []dto.Message
+	for _, content := range geminiRequest.Contents {
+		message := dto.Message{
+			Role: convertGeminiRoleToOpenAI(content.Role),
+		}
+		var mediaContents []dto.MediaContent
+		var toolCalls []dto.ToolCallRequest
+		for _, part := range content.Parts {
+			if part.Text != "" {
+				mediaContent := dto.MediaContent{
+					Type: "text",
+					Text: part.Text,
+				}
+				mediaContents = append(mediaContents, mediaContent)
+			} else if part.InlineData != nil {
+				mediaContent := dto.MediaContent{
+					Type: "image_url",
+					ImageUrl: &dto.MessageImageUrl{
+						Url:      fmt.Sprintf("data:%s;base64,%s", part.InlineData.MimeType, part.InlineData.Data),
+						Detail:   "auto",
+						MimeType: part.InlineData.MimeType,
+					},
+				}
+				mediaContents = append(mediaContents, mediaContent)
+			} else if part.FileData != nil {
+				mediaContent := dto.MediaContent{
+					Type: "image_url",
+					ImageUrl: &dto.MessageImageUrl{
+						Url:      part.FileData.FileUri,
+						Detail:   "auto",
+						MimeType: part.FileData.MimeType,
+					},
+				}
+				mediaContents = append(mediaContents, mediaContent)
+			} else if part.FunctionCall != nil {
+				toolCall := dto.ToolCallRequest{
+					ID:   fmt.Sprintf("call_%d", len(toolCalls)+1),
+					Type: "function",
+					Function: dto.FunctionRequest{
+						Name:      part.FunctionCall.FunctionName,
+						Arguments: jsonutil.ToJSONString(part.FunctionCall.Arguments),
+					},
+				}
+				toolCalls = append(toolCalls, toolCall)
+			} else if part.FunctionResponse != nil {
+				toolMessage := dto.Message{
+					Role:       "tool",
+					ToolCallId: fmt.Sprintf("call_%d", len(toolCalls)),
+				}
+				toolMessage.SetStringContent(jsonutil.ToJSONString(part.FunctionResponse.Response))
+				messages = append(messages, toolMessage)
+			}
+		}
+		if len(toolCalls) > 0 {
+			message.SetToolCalls(toolCalls)
+		} else if len(mediaContents) == 1 && mediaContents[0].Type == "text" {
+			message.Content = mediaContents[0].Text
+		} else if len(mediaContents) > 0 {
+			message.SetMediaContent(mediaContents)
+		}
+		if len(message.ParseContent()) > 0 || len(message.ToolCalls) > 0 {
+			messages = append(messages, message)
+		}
+	}
+	openaiRequest.Messages = messages
+	if geminiRequest.GenerationConfig.Temperature != nil {
+		openaiRequest.Temperature = geminiRequest.GenerationConfig.Temperature
+	}
+	if geminiRequest.GenerationConfig.TopP != nil && *geminiRequest.GenerationConfig.TopP > 0 {
+		openaiRequest.TopP = common.GetPointer(*geminiRequest.GenerationConfig.TopP)
+	}
+	if geminiRequest.GenerationConfig.TopK != nil && *geminiRequest.GenerationConfig.TopK > 0 {
+		openaiRequest.TopK = common.GetPointer(int(*geminiRequest.GenerationConfig.TopK))
+	}
+	if geminiRequest.GenerationConfig.MaxOutputTokens != nil && *geminiRequest.GenerationConfig.MaxOutputTokens > 0 {
+		openaiRequest.MaxTokens = common.GetPointer(*geminiRequest.GenerationConfig.MaxOutputTokens)
+	}
+	if len(geminiRequest.GenerationConfig.StopSequences) > 0 {
+		openaiRequest.Stop = geminiRequest.GenerationConfig.StopSequences[:min(len(geminiRequest.GenerationConfig.StopSequences), 4)]
+	}
+	if geminiRequest.GenerationConfig.CandidateCount != nil && *geminiRequest.GenerationConfig.CandidateCount > 0 {
+		openaiRequest.N = common.GetPointer(*geminiRequest.GenerationConfig.CandidateCount)
+	}
+	if len(geminiRequest.GetTools()) > 0 {
+		var tools []dto.ToolCallRequest
+		for _, tool := range geminiRequest.GetTools() {
+			if tool.FunctionDeclarations == nil {
+				continue
+			}
+			functionDeclarations, err := common.Any2Type[[]dto.FunctionRequest](tool.FunctionDeclarations)
+			if err != nil {
+				common.SysError(fmt.Sprintf("failed to parse gemini function declarations: %v (type=%T)", err, tool.FunctionDeclarations))
+				continue
+			}
+			for _, function := range functionDeclarations {
+				openAITool := dto.ToolCallRequest{
+					Type: "function",
+					Function: dto.FunctionRequest{
+						Name:        function.Name,
+						Description: function.Description,
+						Parameters:  function.Parameters,
+					},
+				}
+				tools = append(tools, openAITool)
+			}
+		}
+		if len(tools) > 0 {
+			openaiRequest.Tools = tools
+		}
+	}
+	if geminiRequest.SystemInstructions != nil {
+		systemMessage := dto.Message{
+			Role:    "system",
+			Content: extractTextFromGeminiParts(geminiRequest.SystemInstructions.Parts),
+		}
+		openaiRequest.Messages = append([]dto.Message{systemMessage}, openaiRequest.Messages...)
+	}
+	return openaiRequest, nil
+}
+func convertGeminiRoleToOpenAI(geminiRole string) string {
+	switch geminiRole {
+	case "user":
+		return "user"
+	case "model":
+		return "assistant"
+	case "function":
+		return "function"
+	default:
+		return "user"
+	}
+}
+func extractTextFromGeminiParts(parts []dto.GeminiPart) string {
+	texts := make([]string, 0)
+	for _, part := range parts {
+		if part.Text != "" {
+			texts = append(texts, part.Text)
+		}
+	}
+	return strings.Join(texts, "\n")
+}
--- a/service/relayconvert/internal/gemini_chat/to_oai_chat_resp.go
+++ b/service/relayconvert/internal/gemini_chat/to_oai_chat_resp.go
+package geminichat
+import (
+	"fmt"
+	"strings"
+	"github.com/QuantumNous/new-api/common"
+	"github.com/QuantumNous/new-api/constant"
+	"github.com/QuantumNous/new-api/dto"
+)
+func UsageFromGeminiMetadata(metadata *dto.GeminiUsageMetadata, fallbackPromptTokens int) *dto.Usage {
+	if metadata == nil {
+		if fallbackPromptTokens <= 0 {
+			return nil
+		}
+		usage := &dto.Usage{PromptTokens: fallbackPromptTokens}
+		usage.PromptTokensDetails.TextTokens = fallbackPromptTokens
+		return usage
+	}
+	promptTokens := metadata.PromptTokenCount + metadata.ToolUsePromptTokenCount
+	if promptTokens <= 0 && fallbackPromptTokens > 0 {
+		promptTokens = fallbackPromptTokens
+	}
+	usage := &dto.Usage{
+		PromptTokens:     promptTokens,
+		CompletionTokens: metadata.CandidatesTokenCount + metadata.ThoughtsTokenCount,
+		TotalTokens:      metadata.TotalTokenCount,
+		BillingUsage:     dto.CloneBillingUsage(metadata.BillingUsage),
+	}
+	if usage.BillingUsage == nil {
+		usage.BillingUsage = dto.NewGeminiChatBillingUsage(metadata)
+	}
+	usage.CompletionTokenDetails.ReasoningTokens = metadata.ThoughtsTokenCount
+	usage.PromptTokensDetails.CachedTokens = metadata.CachedContentTokenCount
+	for _, detail := range metadata.PromptTokensDetails {
+		if detail.Modality == "AUDIO" {
+			usage.PromptTokensDetails.AudioTokens += detail.TokenCount
+		} else if detail.Modality == "IMAGE" {
+			usage.PromptTokensDetails.ImageTokens += detail.TokenCount
+		} else if detail.Modality == "TEXT" {
+			usage.PromptTokensDetails.TextTokens += detail.TokenCount
+		}
+	}
+	for _, detail := range metadata.ToolUsePromptTokensDetails {
+		if detail.Modality == "AUDIO" {
+			usage.PromptTokensDetails.AudioTokens += detail.TokenCount
+		} else if detail.Modality == "IMAGE" {
+			usage.PromptTokensDetails.ImageTokens += detail.TokenCount
+		} else if detail.Modality == "TEXT" {
+			usage.PromptTokensDetails.TextTokens += detail.TokenCount
+		}
+	}
+	for _, detail := range metadata.CandidatesTokensDetails {
+		switch detail.Modality {
+		case "IMAGE":
+			usage.CompletionTokenDetails.ImageTokens += detail.TokenCount
+		case "AUDIO":
+			usage.CompletionTokenDetails.AudioTokens += detail.TokenCount
+		case "TEXT":
+			usage.CompletionTokenDetails.TextTokens += detail.TokenCount
+		}
+	}
+	if usage.TotalTokens > 0 && usage.CompletionTokens <= 0 {
+		usage.CompletionTokens = usage.TotalTokens - usage.PromptTokens
+	}
+	if usage.PromptTokens > 0 && usage.PromptTokensDetails.TextTokens == 0 && usage.PromptTokensDetails.AudioTokens == 0 {
+		usage.PromptTokensDetails.TextTokens = usage.PromptTokens
+	}
+	return usage
+}
+func ResponseGeminiChat2OpenAI(id string, created int64, response *dto.GeminiChatResponse) *dto.OpenAITextResponse {
+	fullTextResponse := dto.OpenAITextResponse{
+		Id:      id,
+		Object:  "chat.completion",
+		Created: created,
+		Choices: make([]dto.OpenAITextResponseChoice, 0, len(response.Candidates)),
+	}
+	isToolCall := false
+	for _, candidate := range response.Candidates {
+		choice := dto.OpenAITextResponseChoice{
+			Index: int(candidate.Index),
+			Message: dto.Message{
+				Role:    "assistant",
+				Content: "",
+			},
+			FinishReason: constant.FinishReasonStop,
+		}
+		if len(candidate.Content.Parts) > 0 {
+			var content strings.Builder
+			var inlineGrow int
+			for _, part := range candidate.Content.Parts {
+				if part.InlineData != nil {
+					inlineGrow += len(part.InlineData.MimeType) + len(part.InlineData.Data) + 32
+				}
+			}
+			if inlineGrow > 0 {
+				content.Grow(inlineGrow)
+			}
+			appended := 0
+			writeSep := func() {
+				if appended > 0 {
+					content.WriteByte('\n')
+				}
+				appended++
+			}
+			var toolCalls []dto.ToolCallResponse
+			for _, part := range candidate.Content.Parts {
+				if part.InlineData != nil {
+					if strings.HasPrefix(part.InlineData.MimeType, "image") {
+						writeSep()
+						content.WriteString("![image](data:")
+						content.WriteString(part.InlineData.MimeType)
+						content.WriteString(";base64,")
+						content.WriteString(part.InlineData.Data)
+						content.WriteByte(')')
+					} else {
+						writeSep()
+						content.WriteString("[media](data:")
+						content.WriteString(part.InlineData.MimeType)
+						content.WriteString(";base64,")
+						content.WriteString(part.InlineData.Data)
+						content.WriteByte(')')
+					}
+				} else if part.FunctionCall != nil {
+					choice.FinishReason = constant.FinishReasonToolCalls
+					if call := geminiResponseToolCall(&part); call != nil {
+						toolCalls = append(toolCalls, *call)
+					}
+				} else if part.Thought {
+					choice.Message.ReasoningContent = &part.Text
+				} else {
+					if part.ExecutableCode != nil {
+						writeSep()
+						content.WriteString("```")
+						content.WriteString(part.ExecutableCode.Language)
+						content.WriteByte('\n')
+						content.WriteString(part.ExecutableCode.Code)
+						content.WriteString("\n```")
+					} else if part.CodeExecutionResult != nil {
+						writeSep()
+						content.WriteString("```output\n")
+						content.WriteString(part.CodeExecutionResult.Output)
+						content.WriteString("\n```")
+					} else if part.Text != "\n" {
+						writeSep()
+						content.WriteString(part.Text)
+					}
+				}
+			}
+			if len(toolCalls) > 0 {
+				choice.Message.SetToolCalls(toolCalls)
+				isToolCall = true
+			}
+			choice.Message.SetStringContent(content.String())
+		}
+		if candidate.FinishReason != nil {
+			switch *candidate.FinishReason {
+			case "STOP":
+				choice.FinishReason = constant.FinishReasonStop
+			case "MAX_TOKENS":
+				choice.FinishReason = constant.FinishReasonLength
+			case "SAFETY", "RECITATION", "BLOCKLIST", "PROHIBITED_CONTENT", "SPII", "OTHER":
+				choice.FinishReason = constant.FinishReasonContentFilter
+			default:
+				choice.FinishReason = constant.FinishReasonContentFilter
+			}
+		}
+		if isToolCall {
+			choice.FinishReason = constant.FinishReasonToolCalls
+		}
+		fullTextResponse.Choices = append(fullTextResponse.Choices, choice)
+	}
+	return &fullTextResponse
+}
+func StreamResponseGeminiChat2OpenAI(geminiResponse *dto.GeminiChatResponse) (*dto.ChatCompletionsStreamResponse, bool) {
+	choices := make([]dto.ChatCompletionsStreamResponseChoice, 0, len(geminiResponse.Candidates))
+	isStop := false
+	for _, candidate := range geminiResponse.Candidates {
+		if candidate.FinishReason != nil && *candidate.FinishReason == "STOP" {
+			isStop = true
+			candidate.FinishReason = nil
+		}
+		choice := dto.ChatCompletionsStreamResponseChoice{
+			Index: int(candidate.Index),
+			Delta: dto.ChatCompletionsStreamResponseChoiceDelta{},
+		}
+		var content strings.Builder
+		var inlineGrow int
+		for _, part := range candidate.Content.Parts {
+			if part.InlineData != nil {
+				inlineGrow += len(part.InlineData.MimeType) + len(part.InlineData.Data) + 32
+			}
+		}
+		if inlineGrow > 0 {
+			content.Grow(inlineGrow)
+		}
+		appended := 0
+		writeSep := func() {
+			if appended > 0 {
+				content.WriteByte('\n')
+			}
+			appended++
+		}
+		isTools := false
+		isThought := false
+		if candidate.FinishReason != nil {
+			switch *candidate.FinishReason {
+			case "STOP":
+				choice.FinishReason = &constant.FinishReasonStop
+			case "MAX_TOKENS":
+				choice.FinishReason = &constant.FinishReasonLength
+			case "SAFETY", "RECITATION", "BLOCKLIST", "PROHIBITED_CONTENT", "SPII", "OTHER":
+				choice.FinishReason = &constant.FinishReasonContentFilter
+			default:
+				choice.FinishReason = &constant.FinishReasonContentFilter
+			}
+		}
+		for _, part := range candidate.Content.Parts {
+			if part.InlineData != nil {
+				if strings.HasPrefix(part.InlineData.MimeType, "image") {
+					writeSep()
+					content.WriteString("![image](data:")
+					content.WriteString(part.InlineData.MimeType)
+					content.WriteString(";base64,")
+					content.WriteString(part.InlineData.Data)
+					content.WriteByte(')')
+				}
+			} else if part.FunctionCall != nil {
+				isTools = true
+				if call := geminiResponseToolCall(&part); call != nil {
+					call.SetIndex(len(choice.Delta.ToolCalls))
+					choice.Delta.ToolCalls = append(choice.Delta.ToolCalls, *call)
+				}
+			} else if part.Thought {
+				isThought = true
+				writeSep()
+				content.WriteString(part.Text)
+			} else {
+				if part.ExecutableCode != nil {
+					writeSep()
+					content.WriteString("```")
+					content.WriteString(part.ExecutableCode.Language)
+					content.WriteByte('\n')
+					content.WriteString(part.ExecutableCode.Code)
+					content.WriteString("\n```\n")
+				} else if part.CodeExecutionResult != nil {
+					writeSep()
+					content.WriteString("```output\n")
+					content.WriteString(part.CodeExecutionResult.Output)
+					content.WriteString("\n```\n")
+				} else if part.Text != "\n" {
+					writeSep()
+					content.WriteString(part.Text)
+				}
+			}
+		}
+		if isThought {
+			choice.Delta.SetReasoningContent(content.String())
+		} else {
+			choice.Delta.SetContentString(content.String())
+		}
+		if isTools {
+			choice.FinishReason = &constant.FinishReasonToolCalls
+		}
+		choices = append(choices, choice)
+	}
+	response := dto.ChatCompletionsStreamResponse{
+		Object:  "chat.completion.chunk",
+		Choices: choices,
+	}
+	return &response, isStop
+}
+func geminiResponseToolCall(item *dto.GeminiPart) *dto.ToolCallResponse {
+	argsBytes, err := common.Marshal(item.FunctionCall.Arguments)
+	if err != nil {
+		return nil
+	}
+	return &dto.ToolCallResponse{
+		ID:   fmt.Sprintf("call_%s", common.GetUUID()),
+		Type: "function",
+		Function: dto.FunctionResponse{
+			Arguments: string(argsBytes),
+			Name:      item.FunctionCall.FunctionName,
+		},
+	}
+}
--- a/service/relayconvert/internal/jsonutil/stringify.go
+++ b/service/relayconvert/internal/jsonutil/stringify.go
+package jsonutil
+import (
+	"fmt"
+	"github.com/QuantumNous/new-api/common"
+)
+func ToJSONString(v interface{}) string {
+	bytes, err := common.Marshal(v)
+	if err != nil {
+		return fmt.Sprintf("%v", v)
+	}
+	return string(bytes)
+}
--- a/service/relayconvert/regex.go
+++ b/service/relayconvert/regex.go
-package relayconvert
+package matcher
 import (
 	"regexp"
@@ -7,7 +7,7 @@ import (
 var compiledRegexCache sync.Map // map[string]*regexp.Regexp
-func matchAnyRegex(patterns []string, s string) bool {
+func MatchAnyRegex(patterns []string, s string) bool {
 	if len(patterns) == 0 || s == "" {
 		return false
 	}

--- a/service/relayconvert/internal/media/media.go
+++ b/service/relayconvert/internal/media/media.go
+package media
+import (
+	"errors"
+	"sync"
+	"github.com/QuantumNous/new-api/types"
+	"github.com/gin-gonic/gin"
+)
+type MediaResolver struct {
+	GetBase64Data        func(c *gin.Context, source types.FileSource, reason ...string) (string, string, error)
+	DecodeBase64FileData func(base64String string) (string, string, error)
+}
+var (
+	mediaResolverMu sync.RWMutex
+	mediaResolver   MediaResolver
+)
+func SetMediaResolver(resolver MediaResolver) {
+	mediaResolverMu.Lock()
+	defer mediaResolverMu.Unlock()
+	mediaResolver = resolver
+}
+func ResolveBase64Data(c *gin.Context, source types.FileSource, reason ...string) (string, string, error) {
+	mediaResolverMu.RLock()
+	resolver := mediaResolver.GetBase64Data
+	mediaResolverMu.RUnlock()
+	if resolver == nil {
+		return "", "", errors.New("relayconvert media resolver is not configured")
+	}
+	return resolver(c, source, reason...)
+}
+func DecodeBase64FileData(base64String string) (string, string, error) {
+	mediaResolverMu.RLock()
+	resolver := mediaResolver.DecodeBase64FileData
+	mediaResolverMu.RUnlock()
+	if resolver == nil {
+		return "", "", errors.New("relayconvert media resolver is not configured")
+	}
+	return resolver(base64String)
+}
--- a/service/relayconvert/internal/meta/relay_info.go
+++ b/service/relayconvert/internal/meta/relay_info.go
+package meta
+import relaycommon "github.com/QuantumNous/new-api/relay/common"
+func RelayInfoChannelType(info *relaycommon.RelayInfo) int {
+	if info == nil || info.ChannelMeta == nil {
+		return 0
+	}
+	return info.ChannelType
+}
+func RelayInfoUpstreamModelName(info *relaycommon.RelayInfo) string {
+	if info == nil || info.ChannelMeta == nil {
+		return ""
+	}
+	return info.UpstreamModelName
+}
--- a/service/relayconvert/internal/oai_chat/to_claude_messages_req.go
+++ b/service/relayconvert/internal/oai_chat/to_claude_messages_req.go
--- a/service/relayconvert/internal/oai_chat/to_claude_messages_resp.go
+++ b/service/relayconvert/internal/oai_chat/to_claude_messages_resp.go
--- a/service/relayconvert/internal/oai_chat/to_claude_messages_resp_test.go
+++ b/service/relayconvert/internal/oai_chat/to_claude_messages_resp_test.go
+package oaichat
+import (
+	"testing"
+	"github.com/QuantumNous/new-api/dto"
+	relaycommon "github.com/QuantumNous/new-api/relay/common"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+func TestResponseOpenAI2ClaudeToolUseInputIsObject(t *testing.T) {
+	tests := []struct {
+		name string
+		args string
+		want map[string]interface{}
+	}{
+		{name: "object", args: `{"q":"x"}`, want: map[string]interface{}{"q": "x"}},
+		{name: "empty", args: "", want: map[string]interface{}{}},
+		{name: "invalid", args: "{", want: map[string]interface{}{}},
+		{name: "null", args: "null", want: map[string]interface{}{}},
+		{name: "array", args: `["x"]`, want: map[string]interface{}{}},
+		{name: "string", args: `"x"`, want: map[string]interface{}{}},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			msg := dto.Message{Role: "assistant"}
+			msg.SetToolCalls([]dto.ToolCallRequest{
+				{
+					ID:   "call_1",
+					Type: "function",
+					Function: dto.FunctionRequest{
+						Name:      "lookup",
+						Arguments: tt.args,
+					},
+				},
+			})
+			resp := ResponseOpenAI2Claude(&dto.OpenAITextResponse{
+				Id:    "chatcmpl_1",
+				Model: "gpt-test",
+				Choices: []dto.OpenAITextResponseChoice{
+					{Message: msg, FinishReason: "tool_calls"},
+				},
+			}, nil)
+			require.Len(t, resp.Content, 1)
+			assert.Equal(t, "tool_use", resp.Content[0].Type)
+			assert.Equal(t, tt.want, resp.Content[0].Input)
+		})
+	}
+}
+func TestResponseOpenAI2ClaudeUsageCarriesOpenAIBillingUsage(t *testing.T) {
+	resp := ResponseOpenAI2Claude(&dto.OpenAITextResponse{
+		Id:    "chatcmpl_1",
+		Model: "gpt-test",
+		Choices: []dto.OpenAITextResponseChoice{
+			{Message: dto.Message{Role: "assistant", Content: "hello"}, FinishReason: "stop"},
+		},
+		Usage: dto.Usage{
+			PromptTokens:     11,
+			CompletionTokens: 5,
+			TotalTokens:      16,
+		},
+	}, nil)
+	require.NotNil(t, resp.Usage)
+	assert.Equal(t, 11, resp.Usage.InputTokens)
+	assert.Equal(t, 5, resp.Usage.OutputTokens)
+	require.NotNil(t, resp.Usage.BillingUsage)
+	require.NotNil(t, resp.Usage.BillingUsage.OpenAIUsage)
+	assert.Equal(t, dto.BillingUsageSourceOAIChat, resp.Usage.BillingUsage.Source)
+	assert.Equal(t, dto.BillingUsageSemanticOpenAI, resp.Usage.BillingUsage.Semantic)
+	assert.Equal(t, 11, resp.Usage.BillingUsage.OpenAIUsage.PromptTokens)
+	assert.Equal(t, 5, resp.Usage.BillingUsage.OpenAIUsage.CompletionTokens)
+	assert.Equal(t, 16, resp.Usage.BillingUsage.OpenAIUsage.TotalTokens)
+	assert.Nil(t, resp.Usage.BillingUsage.OpenAIUsage.BillingUsage)
+}
+func TestStreamResponseOpenAI2ClaudeClosesTextThinkingAndToolBlocks(t *testing.T) {
+	info := &relaycommon.RelayInfo{
+		ClaudeConvertInfo: &relaycommon.ClaudeConvertInfo{
+			LastMessagesType: relaycommon.LastMessageTypeNone,
+		},
+	}
+	info.SendResponseCount = 1
+	textResponses := StreamResponseOpenAI2Claude(&dto.ChatCompletionsStreamResponse{
+		Id:    "chatcmpl_1",
+		Model: "gpt-test",
+		Choices: []dto.ChatCompletionsStreamResponseChoice{
+			{
+				Delta: dto.ChatCompletionsStreamResponseChoiceDelta{
+					Content: ptr("hello"),
+				},
+			},
+		},
+	}, info)
+	require.Len(t, textResponses, 3)
+	assert.Equal(t, "message_start", textResponses[0].Type)
+	assert.Equal(t, "content_block_start", textResponses[1].Type)
+	assert.Equal(t, 0, textResponses[1].GetIndex())
+	assert.Equal(t, "content_block_delta", textResponses[2].Type)
+	info.SendResponseCount = 2
+	thinkingResponses := StreamResponseOpenAI2Claude(&dto.ChatCompletionsStreamResponse{
+		Id:    "chatcmpl_1",
+		Model: "gpt-test",
+		Choices: []dto.ChatCompletionsStreamResponseChoice{
+			{
+				Delta: dto.ChatCompletionsStreamResponseChoiceDelta{
+					ReasoningContent: ptr("thinking"),
+				},
+			},
+		},
+	}, info)
+	require.Len(t, thinkingResponses, 3)
+	assert.Equal(t, "content_block_stop", thinkingResponses[0].Type)
+	assert.Equal(t, 0, thinkingResponses[0].GetIndex())
+	assert.Equal(t, "content_block_start", thinkingResponses[1].Type)
+	assert.Equal(t, 1, thinkingResponses[1].GetIndex())
+	assert.Equal(t, "thinking", thinkingResponses[1].ContentBlock.Type)
+	assert.Equal(t, "content_block_delta", thinkingResponses[2].Type)
+	info.SendResponseCount = 3
+	toolResponses := StreamResponseOpenAI2Claude(&dto.ChatCompletionsStreamResponse{
+		Id:    "chatcmpl_1",
+		Model: "gpt-test",
+		Choices: []dto.ChatCompletionsStreamResponseChoice{
+			{
+				Delta: dto.ChatCompletionsStreamResponseChoiceDelta{
+					ToolCalls: []dto.ToolCallResponse{
+						{
+							Index: ptr(0),
+							ID:    "call_1",
+							Type:  "function",
+							Function: dto.FunctionResponse{
+								Name:      "lookup",
+								Arguments: `{"q":"x"}`,
+							},
+						},
+					},
+				},
+			},
+		},
+	}, info)
+	require.Len(t, toolResponses, 3)
+	assert.Equal(t, "content_block_stop", toolResponses[0].Type)
+	assert.Equal(t, 1, toolResponses[0].GetIndex())
+	assert.Equal(t, "content_block_start", toolResponses[1].Type)
+	assert.Equal(t, 2, toolResponses[1].GetIndex())
+	assert.Equal(t, "tool_use", toolResponses[1].ContentBlock.Type)
+	assert.Equal(t, "content_block_delta", toolResponses[2].Type)
+	info.SendResponseCount = 4
+	finishResponses := StreamResponseOpenAI2Claude(&dto.ChatCompletionsStreamResponse{
+		Id:    "chatcmpl_1",
+		Model: "gpt-test",
+		Choices: []dto.ChatCompletionsStreamResponseChoice{
+			{FinishReason: ptr("tool_calls")},
+		},
+		Usage: &dto.Usage{
+			PromptTokens:     7,
+			CompletionTokens: 3,
+			TotalTokens:      10,
+		},
+	}, info)
+	require.Len(t, finishResponses, 3)
+	assert.Equal(t, "content_block_stop", finishResponses[0].Type)
+	assert.Equal(t, 2, finishResponses[0].GetIndex())
+	assert.Equal(t, "message_delta", finishResponses[1].Type)
+	assert.Equal(t, "tool_use", *finishResponses[1].Delta.StopReason)
+	require.NotNil(t, finishResponses[1].Usage)
+	require.NotNil(t, finishResponses[1].Usage.BillingUsage)
+	require.NotNil(t, finishResponses[1].Usage.BillingUsage.OpenAIUsage)
+	assert.Equal(t, 7, finishResponses[1].Usage.BillingUsage.OpenAIUsage.PromptTokens)
+	assert.Equal(t, 3, finishResponses[1].Usage.BillingUsage.OpenAIUsage.CompletionTokens)
+	assert.Equal(t, "message_stop", finishResponses[2].Type)
+}
+func TestNormalizeCacheCreationSplit(t *testing.T) {
+	cache5m, cache1h := NormalizeCacheCreationSplit(10, 3, 2)
+	assert.Equal(t, 8, cache5m)
+	assert.Equal(t, 2, cache1h)
+	cache5m, cache1h = NormalizeCacheCreationSplit(3, 5, 1)
+	assert.Equal(t, 5, cache5m)
+	assert.Equal(t, 1, cache1h)
+}
+func ptr[T any](value T) *T {
+	return &value
+}
--- a/service/relayconvert/internal/oai_chat/to_gemini_chat_req.go
+++ b/service/relayconvert/internal/oai_chat/to_gemini_chat_req.go
--- a/service/relayconvert/internal/oai_chat/to_gemini_chat_resp.go
+++ b/service/relayconvert/internal/oai_chat/to_gemini_chat_resp.go
+package oaichat
+import (
+	"github.com/QuantumNous/new-api/common"
+	"github.com/QuantumNous/new-api/dto"
+	relaycommon "github.com/QuantumNous/new-api/relay/common"
+)
+// ResponseOpenAI2Gemini 将 OpenAI 响应转换为 Gemini 格式
+func ResponseOpenAI2Gemini(openAIResponse *dto.OpenAITextResponse, info *relaycommon.RelayInfo) *dto.GeminiChatResponse {
+	totalTokens := openAIResponse.TotalTokens
+	if totalTokens == 0 {
+		totalTokens = openAIResponse.PromptTokens + openAIResponse.CompletionTokens
+	}
+	geminiResponse := &dto.GeminiChatResponse{
+		Candidates:       make([]dto.GeminiChatCandidate, 0, len(openAIResponse.Choices)),
+		HasUsageMetadata: true,
+		UsageMetadata: dto.GeminiUsageMetadata{
+			PromptTokenCount:     openAIResponse.PromptTokens,
+			CandidatesTokenCount: openAIResponse.CompletionTokens,
+			TotalTokenCount:      totalTokens,
+			BillingUsage:         openAIBillingUsageFromUsage(&openAIResponse.Usage),
+		},
+	}
+	if metadata, ok := geminiBillingMetadataFromOpenAIUsage(&openAIResponse.Usage); ok {
+		geminiResponse.UsageMetadata = metadata
+	}
+	for _, choice := range openAIResponse.Choices {
+		candidate := dto.GeminiChatCandidate{
+			Index:         int64(choice.Index),
+			SafetyRatings: []dto.GeminiChatSafetyRating{},
+		}
+		// 设置结束原因
+		var finishReason string
+		switch choice.FinishReason {
+		case "stop":
+			finishReason = "STOP"
+		case "length":
+			finishReason = "MAX_TOKENS"
+		case "content_filter":
+			finishReason = "SAFETY"
+		case "tool_calls":
+			finishReason = "STOP"
+		default:
+			finishReason = "STOP"
+		}
+		candidate.FinishReason = &finishReason
+		// 转换消息内容
+		content := dto.GeminiChatContent{
+			Role:  "model",
+			Parts: make([]dto.GeminiPart, 0),
+		}
+		textContent := choice.Message.StringContent()
+		if textContent != "" {
+			part := dto.GeminiPart{
+				Text: textContent,
+			}
+			content.Parts = append(content.Parts, part)
+		}
+		toolCalls := choice.Message.ParseToolCalls()
+		for _, toolCall := range toolCalls {
+			var args map[string]interface{}
+			if toolCall.Function.Arguments != "" {
+				if err := common.Unmarshal([]byte(toolCall.Function.Arguments), &args); err != nil {
+					args = map[string]interface{}{"arguments": toolCall.Function.Arguments}
+				}
+			} else {
+				args = make(map[string]interface{})
+			}
+			part := dto.GeminiPart{
+				FunctionCall: &dto.FunctionCall{
+					FunctionName: toolCall.Function.Name,
+					Arguments:    args,
+				},
+			}
+			content.Parts = append(content.Parts, part)
+		}
+		candidate.Content = content
+		geminiResponse.Candidates = append(geminiResponse.Candidates, candidate)
+	}
+	return geminiResponse
+}
+// StreamResponseOpenAI2Gemini 将 OpenAI 流式响应转换为 Gemini 格式
+func StreamResponseOpenAI2Gemini(openAIResponse *dto.ChatCompletionsStreamResponse, info *relaycommon.RelayInfo) *dto.GeminiChatResponse {
+	// 检查是否有实际内容或结束标志
+	hasContent := false
+	hasFinishReason := false
+	for _, choice := range openAIResponse.Choices {
+		if len(choice.Delta.GetContentString()) > 0 || (choice.Delta.ToolCalls != nil && len(choice.Delta.ToolCalls) > 0) {
+			hasContent = true
+		}
+		if choice.FinishReason != nil {
+			hasFinishReason = true
+		}
+	}
+	// 如果没有实际内容且没有结束标志，跳过。主要针对 openai 流响应开头的空数据
+	if !hasContent && !hasFinishReason {
+		return nil
+	}
+	estimatePromptTokens := 0
+	if info != nil {
+		estimatePromptTokens = info.GetEstimatePromptTokens()
+	}
+	geminiResponse := &dto.GeminiChatResponse{
+		Candidates:       make([]dto.GeminiChatCandidate, 0, len(openAIResponse.Choices)),
+		HasUsageMetadata: true,
+		UsageMetadata: dto.GeminiUsageMetadata{
+			PromptTokenCount:     estimatePromptTokens,
+			CandidatesTokenCount: 0, // 流式响应中可能没有完整的 usage 信息
+			TotalTokenCount:      estimatePromptTokens,
+		},
+	}
+	if openAIResponse.Usage != nil {
+		geminiResponse.UsageMetadata.PromptTokenCount = openAIResponse.Usage.PromptTokens
+		geminiResponse.UsageMetadata.CandidatesTokenCount = openAIResponse.Usage.CompletionTokens
+		geminiResponse.UsageMetadata.TotalTokenCount = openAIResponse.Usage.TotalTokens
+		geminiResponse.UsageMetadata.BillingUsage = openAIBillingUsageFromUsage(openAIResponse.Usage)
+		if metadata, ok := geminiBillingMetadataFromOpenAIUsage(openAIResponse.Usage); ok {
+			geminiResponse.UsageMetadata = metadata
+		}
+	}
+	for _, choice := range openAIResponse.Choices {
+		candidate := dto.GeminiChatCandidate{
+			Index:         int64(choice.Index),
+			SafetyRatings: []dto.GeminiChatSafetyRating{},
+		}
+		// 设置结束原因
+		if choice.FinishReason != nil {
+			var finishReason string
+			switch *choice.FinishReason {
+			case "stop":
+				finishReason = "STOP"
+			case "length":
+				finishReason = "MAX_TOKENS"
+			case "content_filter":
+				finishReason = "SAFETY"
+			case "tool_calls":
+				finishReason = "STOP"
+			default:
+				finishReason = "STOP"
+			}
+			candidate.FinishReason = &finishReason
+		}
+		// 转换消息内容
+		content := dto.GeminiChatContent{
+			Role:  "model",
+			Parts: make([]dto.GeminiPart, 0),
+		}
+		// 处理工具调用
+		if choice.Delta.ToolCalls != nil {
+			for _, toolCall := range choice.Delta.ToolCalls {
+				// 解析参数
+				var args map[string]interface{}
+				if toolCall.Function.Arguments != "" {
+					if err := common.Unmarshal([]byte(toolCall.Function.Arguments), &args); err != nil {
+						args = map[string]interface{}{"arguments": toolCall.Function.Arguments}
+					}
+				} else {
+					args = make(map[string]interface{})
+				}
+				part := dto.GeminiPart{
+					FunctionCall: &dto.FunctionCall{
+						FunctionName: toolCall.Function.Name,
+						Arguments:    args,
+					},
+				}
+				content.Parts = append(content.Parts, part)
+			}
+		} else {
+			// 处理文本内容
+			textContent := choice.Delta.GetContentString()
+			if textContent != "" {
+				part := dto.GeminiPart{
+					Text: textContent,
+				}
+				content.Parts = append(content.Parts, part)
+			}
+		}
+		candidate.Content = content
+		geminiResponse.Candidates = append(geminiResponse.Candidates, candidate)
+	}
+	return geminiResponse
+}
+func geminiBillingMetadataFromOpenAIUsage(usage *dto.Usage) (dto.GeminiUsageMetadata, bool) {
+	if usage == nil || usage.BillingUsage == nil || usage.BillingUsage.GeminiUsageMetadata == nil {
+		return dto.GeminiUsageMetadata{}, false
+	}
+	if usage.BillingUsage.Source != dto.BillingUsageSourceGeminiChat && usage.BillingUsage.Semantic != dto.BillingUsageSemanticGemini {
+		return dto.GeminiUsageMetadata{}, false
+	}
+	billingUsage := dto.CloneBillingUsage(usage.BillingUsage)
+	if billingUsage == nil || billingUsage.GeminiUsageMetadata == nil {
+		return dto.GeminiUsageMetadata{}, false
+	}
+	return *billingUsage.GeminiUsageMetadata, true
+}
+func openAIBillingUsageFromUsage(usage *dto.Usage) *dto.BillingUsage {
+	if usage == nil {
+		return nil
+	}
+	if existingBillingUsage := dto.CloneBillingUsage(usage.BillingUsage); existingBillingUsage != nil && existingBillingUsage.OpenAIUsage != nil {
+		if existingBillingUsage.Source == dto.BillingUsageSourceOAIChat ||
+			existingBillingUsage.Source == dto.BillingUsageSourceOAIResponses ||
+			existingBillingUsage.Semantic == dto.BillingUsageSemanticOpenAI {
+			return existingBillingUsage
+		}
+	}
+	return dto.NewOpenAIChatBillingUsage(usage)
+}
--- a/service/relayconvert/internal/oai_chat/to_gemini_chat_resp_test.go
+++ b/service/relayconvert/internal/oai_chat/to_gemini_chat_resp_test.go
+package oaichat
+import (
+	"testing"
+	"github.com/QuantumNous/new-api/dto"
+	relaycommon "github.com/QuantumNous/new-api/relay/common"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+func TestResponseOpenAI2GeminiMapsTextToolFinishReasonAndUsage(t *testing.T) {
+	msg := dto.Message{
+		Role:    "assistant",
+		Content: "hello",
+	}
+	msg.SetToolCalls([]dto.ToolCallRequest{
+		{
+			ID:   "call_1",
+			Type: "function",
+			Function: dto.FunctionRequest{
+				Name:      "lookup",
+				Arguments: `{"q":"x"}`,
+			},
+		},
+	})
+	resp := ResponseOpenAI2Gemini(&dto.OpenAITextResponse{
+		Model: "gpt-test",
+		Choices: []dto.OpenAITextResponseChoice{
+			{
+				Index:        2,
+				Message:      msg,
+				FinishReason: "length",
+			},
+		},
+		Usage: dto.Usage{
+			PromptTokens:     11,
+			CompletionTokens: 5,
+			TotalTokens:      16,
+		},
+	}, nil)
+	assert.Equal(t, 11, resp.UsageMetadata.PromptTokenCount)
+	assert.Equal(t, 5, resp.UsageMetadata.CandidatesTokenCount)
+	assert.Equal(t, 16, resp.UsageMetadata.TotalTokenCount)
+	require.NotNil(t, resp.UsageMetadata.BillingUsage)
+	require.NotNil(t, resp.UsageMetadata.BillingUsage.OpenAIUsage)
+	assert.Equal(t, dto.BillingUsageSourceOAIChat, resp.UsageMetadata.BillingUsage.Source)
+	assert.Equal(t, dto.BillingUsageSemanticOpenAI, resp.UsageMetadata.BillingUsage.Semantic)
+	assert.Equal(t, 11, resp.UsageMetadata.BillingUsage.OpenAIUsage.PromptTokens)
+	assert.Equal(t, 5, resp.UsageMetadata.BillingUsage.OpenAIUsage.CompletionTokens)
+	assert.Equal(t, 16, resp.UsageMetadata.BillingUsage.OpenAIUsage.TotalTokens)
+	assert.Nil(t, resp.UsageMetadata.BillingUsage.OpenAIUsage.BillingUsage)
+	require.Len(t, resp.Candidates, 1)
+	assert.Equal(t, int64(2), resp.Candidates[0].Index)
+	require.NotNil(t, resp.Candidates[0].FinishReason)
+	assert.Equal(t, "MAX_TOKENS", *resp.Candidates[0].FinishReason)
+	require.Len(t, resp.Candidates[0].Content.Parts, 2)
+	assert.Equal(t, "hello", resp.Candidates[0].Content.Parts[0].Text)
+	require.NotNil(t, resp.Candidates[0].Content.Parts[1].FunctionCall)
+	assert.Equal(t, "lookup", resp.Candidates[0].Content.Parts[1].FunctionCall.FunctionName)
+	assert.Equal(t, map[string]interface{}{"q": "x"}, resp.Candidates[0].Content.Parts[1].FunctionCall.Arguments)
+}
+func TestStreamResponseOpenAI2GeminiMapsToolCallFinishReasonAndUsage(t *testing.T) {
+	resp := StreamResponseOpenAI2Gemini(&dto.ChatCompletionsStreamResponse{
+		Choices: []dto.ChatCompletionsStreamResponseChoice{
+			{
+				Index:        1,
+				FinishReason: geminiRespPtr("tool_calls"),
+				Delta: dto.ChatCompletionsStreamResponseChoiceDelta{
+					ToolCalls: []dto.ToolCallResponse{
+						{
+							Type: "function",
+							Function: dto.FunctionResponse{
+								Name:      "lookup",
+								Arguments: `{"q":"x"}`,
+							},
+						},
+					},
+				},
+			},
+		},
+		Usage: &dto.Usage{
+			PromptTokens:     13,
+			CompletionTokens: 8,
+			TotalTokens:      21,
+		},
+	}, &relaycommon.RelayInfo{})
+	require.NotNil(t, resp)
+	assert.Equal(t, 13, resp.UsageMetadata.PromptTokenCount)
+	assert.Equal(t, 8, resp.UsageMetadata.CandidatesTokenCount)
+	assert.Equal(t, 21, resp.UsageMetadata.TotalTokenCount)
+	require.NotNil(t, resp.UsageMetadata.BillingUsage)
+	require.NotNil(t, resp.UsageMetadata.BillingUsage.OpenAIUsage)
+	assert.Equal(t, 13, resp.UsageMetadata.BillingUsage.OpenAIUsage.PromptTokens)
+	assert.Equal(t, 8, resp.UsageMetadata.BillingUsage.OpenAIUsage.CompletionTokens)
+	require.Len(t, resp.Candidates, 1)
+	assert.Equal(t, int64(1), resp.Candidates[0].Index)
+	require.NotNil(t, resp.Candidates[0].FinishReason)
+	assert.Equal(t, "STOP", *resp.Candidates[0].FinishReason)
+	require.Len(t, resp.Candidates[0].Content.Parts, 1)
+	require.NotNil(t, resp.Candidates[0].Content.Parts[0].FunctionCall)
+	assert.Equal(t, "lookup", resp.Candidates[0].Content.Parts[0].FunctionCall.FunctionName)
+	assert.Equal(t, map[string]interface{}{"q": "x"}, resp.Candidates[0].Content.Parts[0].FunctionCall.Arguments)
+}
+func geminiRespPtr[T any](value T) *T {
+	return &value
+}
--- a/service/relayconvert/policy.go
+++ b/service/relayconvert/policy.go
-package relayconvert
+package oaichat
-import "github.com/QuantumNous/new-api/setting/model_setting"
+import (
+	"github.com/QuantumNous/new-api/service/relayconvert/internal/matcher"
+	"github.com/QuantumNous/new-api/setting/model_setting"
+)
 func ShouldChatCompletionsUseResponsesPolicy(policy model_setting.ChatCompletionsToResponsesPolicy, channelID int, channelType int, model string) bool {
 	if !policy.IsChannelEnabled(channelID, channelType) {
 		return false
 	}
-	return matchAnyRegex(policy.ModelPatterns, model)
+	return matcher.MatchAnyRegex(policy.ModelPatterns, model)
 }
 func ShouldChatCompletionsUseResponsesGlobal(channelID int, channelType int, model string) bool {

--- a/service/relayconvert/chat_to_responses.go
+++ b/service/relayconvert/chat_to_responses.go
-package relayconvert
+package oaichat
 import (
 	"encoding/json"

--- a/service/relayconvert/internal/oai_chat/to_oai_responses_req_test.go
+++ b/service/relayconvert/internal/oai_chat/to_oai_responses_req_test.go
+package oaichat
+import (
+	"testing"
+	"github.com/QuantumNous/new-api/dto"
+	"github.com/samber/lo"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"github.com/tidwall/gjson"
+)
+func TestChatCompletionsRequestToResponsesRequestInstructionsAndTools(t *testing.T) {
+	req := &dto.GeneralOpenAIRequest{
+		Model: "gpt-test",
+		N:     lo.ToPtr(1),
+		Messages: []dto.Message{
+			{Role: "system", Content: "system rules"},
+			{Role: "developer", Content: "developer rules"},
+			{Role: "user", Content: []any{
+				map[string]any{"type": "text", "text": "look"},
+				map[string]any{"type": "image_url", "image_url": map[string]any{"url": "https://example.test/a.png"}},
+			}},
+			assistantMessageWithTool("partial text", "call_1", "lookup", `{"q":"x"}`),
+			{Role: "tool", ToolCallId: "call_1", Content: "tool result"},
+		},
+	}
+	got, err := ChatCompletionsRequestToResponsesRequest(req)
+	require.NoError(t, err)
+	assert.Equal(t, "gpt-test", got.Model)
+	assert.Equal(t, `"system rules\n\ndeveloper rules"`, string(got.Instructions))
+	assert.Equal(t, "input_image", gjson.GetBytes(got.Input, "0.content.1.type").String())
+	assert.Equal(t, "function_call", gjson.GetBytes(got.Input, "2.type").String())
+	assert.Equal(t, "call_1", gjson.GetBytes(got.Input, "2.call_id").String())
+	assert.Equal(t, "function_call_output", gjson.GetBytes(got.Input, "3.type").String())
+}
+func TestChatCompletionsRequestToResponsesRequestRejectsMultipleChoices(t *testing.T) {
+	_, err := ChatCompletionsRequestToResponsesRequest(&dto.GeneralOpenAIRequest{
+		Model: "gpt-test",
+		N:     lo.ToPtr(2),
+	})
+	require.Error(t, err)
+	assert.Contains(t, err.Error(), "n>1")
+}
+func assistantMessageWithTool(content string, id string, name string, args string) dto.Message {
+	msg := dto.Message{Role: "assistant", Content: content}
+	msg.SetToolCalls([]dto.ToolCallRequest{
+		{
+			ID:   id,
+			Type: "function",
+			Function: dto.FunctionRequest{
+				Name:      name,
+				Arguments: args,
+			},
+		},
+	})
+	return msg
+}
--- a/service/relayconvert/internal/oai_chat/to_oai_responses_resp.go
+++ b/service/relayconvert/internal/oai_chat/to_oai_responses_resp.go
+package oaichat
+import (
+	"errors"
+	"fmt"
+	"strings"
+	"time"
+	"github.com/QuantumNous/new-api/common"
+	"github.com/QuantumNous/new-api/dto"
+)
+const (
+	chatFinishReasonLength        = "length"
+	chatFinishReasonContentFilter = "content_filter"
+	responsesEventCreated                  = "response.created"
+	responsesEventCompleted                = "response.completed"
+	responsesEventIncomplete               = "response.incomplete"
+	responsesEventOutputTextDelta          = "response.output_text.delta"
+	responsesEventOutputItemAdded          = "response.output_item.added"
+	responsesEventOutputItemDone           = "response.output_item.done"
+	responsesEventFunctionArgsDelta        = "response.function_call_arguments.delta"
+	responsesEventFunctionArgsDone         = "response.function_call_arguments.done"
+	responsesEventReasoningSummaryDelta    = "response.reasoning_summary_text.delta"
+	responsesEventReasoningSummaryDone     = "response.reasoning_summary_text.done"
+	responsesOutputTypeFunctionCall        = "function_call"
+	responsesOutputTypeMessage             = "message"
+	responsesOutputTypeReasoning           = "reasoning"
+	responsesIncompleteReasonContentFilter = "content_filter"
+	responsesIncompleteReasonMaxTokens     = "max_output_tokens"
+)
+func ChatCompletionsResponseToResponsesResponse(resp *dto.OpenAITextResponse, id string) (*dto.OpenAIResponsesResponse, *dto.Usage, error) {
+	if resp == nil {
+		return nil, nil, errors.New("response is nil")
+	}
+	usage := UsageFromChatUsage(&resp.Usage)
+	out := &dto.OpenAIResponsesResponse{
+		ID:        id,
+		Object:    "response",
+		CreatedAt: chatCreatedAt(resp.Created),
+		Status:    []byte(`"completed"`),
+		Model:     resp.Model,
+		Output:    make([]dto.ResponsesOutput, 0),
+		Usage:     usage,
+	}
+	if len(resp.Choices) == 0 {
+		return out, usage, nil
+	}
+	choice := resp.Choices[0]
+	if status, details := ResponsesStatusFromChatFinishReason(choice.FinishReason); status != "" {
+		out.Status = []byte(fmt.Sprintf("%q", status))
+		out.IncompleteDetails = details
+	}
+	if text := choice.Message.StringContent(); text != "" {
+		out.Output = append(out.Output, dto.ResponsesOutput{
+			Type:   responsesOutputTypeMessage,
+			ID:     fmt.Sprintf("%s_msg_0", id),
+			Status: responseOutputStatus(out),
+			Role:   "assistant",
+			Content: []dto.ResponsesOutputContent{
+				{
+					Type:        "output_text",
+					Text:        text,
+					Annotations: []interface{}{},
+				},
+			},
+		})
+	}
+	if reasoning := choice.Message.GetReasoningContent(); reasoning != "" {
+		out.Output = append(out.Output, dto.ResponsesOutput{
+			Type:   responsesOutputTypeReasoning,
+			ID:     fmt.Sprintf("%s_reasoning_0", id),
+			Status: responseOutputStatus(out),
+			Content: []dto.ResponsesOutputContent{
+				{
+					Type: "summary_text",
+					Text: reasoning,
+				},
+			},
+		})
+	}
+	for i, toolCall := range choice.Message.ParseToolCalls() {
+		toolOutput, err := chatToolCallToResponsesOutput(toolCall, id, i, responseOutputStatus(out))
+		if err != nil {
+			return nil, nil, err
+		}
+		out.Output = append(out.Output, toolOutput)
+	}
+	return out, usage, nil
+}
+func ResponsesStatusFromChatFinishReason(finishReason string) (string, *dto.IncompleteDetails) {
+	switch strings.TrimSpace(finishReason) {
+	case chatFinishReasonLength:
+		return "incomplete", &dto.IncompleteDetails{Reason: responsesIncompleteReasonMaxTokens}
+	case chatFinishReasonContentFilter:
+		return "incomplete", &dto.IncompleteDetails{Reason: responsesIncompleteReasonContentFilter}
+	default:
+		return "completed", nil
+	}
+}
+func UsageFromChatUsage(src *dto.Usage) *dto.Usage {
+	usage := &dto.Usage{}
+	if src == nil {
+		return usage
+	}
+	usage.UsageSemantic = src.UsageSemantic
+	usage.UsageSource = src.UsageSource
+	usage.BillingUsage = dto.CloneBillingUsage(src.BillingUsage)
+	if usage.BillingUsage == nil {
+		usage.BillingUsage = dto.NewOpenAIChatBillingUsage(src)
+	}
+	usage.Cost = src.Cost
+	if src.PromptTokens != 0 {
+		usage.PromptTokens = src.PromptTokens
+		usage.InputTokens = src.PromptTokens
+	}
+	if src.CompletionTokens != 0 {
+		usage.CompletionTokens = src.CompletionTokens
+		usage.OutputTokens = src.CompletionTokens
+	}
+	if src.TotalTokens != 0 {
+		usage.TotalTokens = src.TotalTokens
+	} else {
+		usage.TotalTokens = usage.InputTokens + usage.OutputTokens
+	}
+	if src.PromptTokensDetails.CachedTokens != 0 ||
+		src.PromptTokensDetails.ImageTokens != 0 ||
+		src.PromptTokensDetails.AudioTokens != 0 ||
+		src.PromptTokensDetails.CachedCreationTokens != 0 ||
+		src.PromptTokensDetails.TextTokens != 0 {
+		details := src.PromptTokensDetails
+		usage.InputTokensDetails = &details
+	}
+	if src.CompletionTokenDetails.ReasoningTokens != 0 ||
+		src.CompletionTokenDetails.TextTokens != 0 ||
+		src.CompletionTokenDetails.AudioTokens != 0 ||
+		src.CompletionTokenDetails.ImageTokens != 0 {
+		usage.CompletionTokenDetails = src.CompletionTokenDetails
+	}
+	usage.ClaudeCacheCreation5mTokens = src.ClaudeCacheCreation5mTokens
+	usage.ClaudeCacheCreation1hTokens = src.ClaudeCacheCreation1hTokens
+	return usage
+}
+func responseOutputStatus(resp *dto.OpenAIResponsesResponse) string {
+	if resp == nil || responseStatusString(resp) != "incomplete" {
+		return "completed"
+	}
+	return "incomplete"
+}
+func responseStatusString(resp *dto.OpenAIResponsesResponse) string {
+	if resp == nil || len(resp.Status) == 0 {
+		return ""
+	}
+	var status string
+	_ = common.Unmarshal(resp.Status, &status)
+	return strings.TrimSpace(status)
+}
+func chatToolCallToResponsesOutput(toolCall dto.ToolCallRequest, responseID string, index int, status string) (dto.ResponsesOutput, error) {
+	callID := strings.TrimSpace(toolCall.ID)
+	if callID == "" {
+		callID = fmt.Sprintf("%s_call_%d", responseID, index)
+	}
+	if toolCall.Type == "" || toolCall.Type == "function" {
+		return dto.ResponsesOutput{
+			Type:      responsesOutputTypeFunctionCall,
+			ID:        callID,
+			Status:    status,
+			CallId:    callID,
+			Name:      toolCall.Function.Name,
+			Arguments: chatArgumentsRawMessage(toolCall.Function.Arguments),
+		}, nil
+	}
+	return dto.ResponsesOutput{
+		Type:      toolCall.Type,
+		ID:        callID,
+		Status:    status,
+		CallId:    callID,
+		Arguments: toolCall.Custom,
+	}, nil
+}
+func chatArgumentsRawMessage(arguments string) []byte {
+	raw, err := common.Marshal(arguments)
+	if err != nil {
+		return []byte(`""`)
+	}
+	return raw
+}
+func chatCreatedAt(created any) int {
+	switch v := created.(type) {
+	case int:
+		return v
+	case int64:
+		return int(v)
+	case float64:
+		return int(v)
+	case float32:
+		return int(v)
+	case string:
+		if parsed := common.String2Int(v); parsed != 0 {
+			return parsed
+		}
+	}
+	return int(time.Now().Unix())
+}
+func responsesStreamEvent(eventType string, payload dto.ResponsesStreamResponse) ChatToResponsesStreamEvent {
+	payload.Type = eventType
+	return ChatToResponsesStreamEvent{
+		Type:    eventType,
+		Payload: payload,
+	}
+}
+func intPtr(v int) *int {
+	return &v
+}
--- a/service/relayconvert/internal/oai_chat/to_oai_responses_resp_test.go
+++ b/service/relayconvert/internal/oai_chat/to_oai_responses_resp_test.go
--- a/service/relayconvert/chat_to_responses_response.go
+++ b/service/relayconvert/chat_to_responses_response.go
--- a/service/relayconvert/internal/oai_responses/req_helpers.go
+++ b/service/relayconvert/internal/oai_responses/req_helpers.go
--- a/service/relayconvert/internal/oai_responses/to_claude_messages_req.go
+++ b/service/relayconvert/internal/oai_responses/to_claude_messages_req.go
--- a/service/relayconvert/internal/oai_responses/to_gemini_chat_req.go
+++ b/service/relayconvert/internal/oai_responses/to_gemini_chat_req.go
--- a/relay/channel/gemini/adaptor_responses.go
+++ b/relay/channel/gemini/adaptor_responses.go
--- a/service/relayconvert/responses_request_to_chat.go
+++ b/service/relayconvert/responses_request_to_chat.go
--- a/service/relayconvert/responses_request_to_chat_test.go
+++ b/service/relayconvert/responses_request_to_chat_test.go
--- a/service/relayconvert/internal/oai_responses/to_oai_chat_resp.go
+++ b/service/relayconvert/internal/oai_responses/to_oai_chat_resp.go
--- a/service/relayconvert/chat_responses_compat_test.go
+++ b/service/relayconvert/chat_responses_compat_test.go
--- a/service/relayconvert/responses_to_chat.go
+++ b/service/relayconvert/responses_to_chat.go
--- a/service/relayconvert/internal/shared/claude/cache.go
+++ b/service/relayconvert/internal/shared/claude/cache.go
--- a/service/relayconvert/internal/shared/claude/tool_choice.go
+++ b/service/relayconvert/internal/shared/claude/tool_choice.go
--- a/service/relayconvert/internal/shared/gemini/request.go
+++ b/service/relayconvert/internal/shared/gemini/request.go
--- a/service/relayconvert/internal/shared/gemini/schema.go
+++ b/service/relayconvert/internal/shared/gemini/schema.go
--- a/service/relayconvert/media.go
+++ b/service/relayconvert/media.go
--- a/service/relayconvert/request_compat.go
+++ b/service/relayconvert/request_compat.go
--- a/service/relayconvert/request_registry.go
+++ b/service/relayconvert/request_registry.go
--- a/service/relayconvert/request_registry_test.go
+++ b/service/relayconvert/request_registry_test.go
--- a/service/relayconvert/response_compat.go
+++ b/service/relayconvert/response_compat.go
--- a/service/relayconvert/response_registry.go
+++ b/service/relayconvert/response_registry.go
--- a/service/relayconvert/response_registry_test.go
+++ b/service/relayconvert/response_registry_test.go
--- a/service/relayconvert/text_converter_registry.go
+++ b/service/relayconvert/text_converter_registry.go
--- a/service/relayconvert/text_converter_registry_test.go
+++ b/service/relayconvert/text_converter_registry_test.go
--- a/service/request_converter.go
+++ b/service/request_converter.go
--- a/service/text_quota.go
+++ b/service/text_quota.go
--- a/service/text_quota_test.go
+++ b/service/text_quota_test.go
--- a/web/default/src/features/channels/components/dialogs/advanced-custom-editor-dialog.tsx
+++ b/web/default/src/features/channels/components/dialogs/advanced-custom-editor-dialog.tsx
--- a/web/default/src/features/channels/lib/advanced-custom.ts
+++ b/web/default/src/features/channels/lib/advanced-custom.ts
--- a/web/default/src/features/channels/types.ts
+++ b/web/default/src/features/channels/types.ts
--- a/web/default/src/features/usage-logs/components/dialogs/details-dialog.tsx
+++ b/web/default/src/features/usage-logs/components/dialogs/details-dialog.tsx
--- a/web/default/src/features/usage-logs/types.ts
+++ b/web/default/src/features/usage-logs/types.ts
--- a/web/default/src/i18n/locales/en.json
+++ b/web/default/src/i18n/locales/en.json
--- a/web/default/src/i18n/locales/fr.json
+++ b/web/default/src/i18n/locales/fr.json
--- a/web/default/src/i18n/locales/ja.json
+++ b/web/default/src/i18n/locales/ja.json
--- a/web/default/src/i18n/locales/ru.json
+++ b/web/default/src/i18n/locales/ru.json
--- a/web/default/src/i18n/locales/vi.json
+++ b/web/default/src/i18n/locales/vi.json
--- a/web/default/src/i18n/locales/zh-TW.json
+++ b/web/default/src/i18n/locales/zh-TW.json
--- a/web/default/src/i18n/locales/zh.json
+++ b/web/default/src/i18n/locales/zh.json