perf: only return quota when it's not zero

controller/relay-text.go

@@ -347,6 +347,7 @@ func relayTextHelper(c *gin.Context, relayMode int) *OpenAIErrorWithStatusCode {
 		isStream = isStream || strings.HasPrefix(resp.Header.Get("Content-Type"), "text/event-stream")
 		isStream = isStream || strings.HasPrefix(resp.Header.Get("Content-Type"), "text/event-stream")
 
 
 		if resp.StatusCode != http.StatusOK {
 		if resp.StatusCode != http.StatusOK {
+			if preConsumedQuota != 0 {
 				go func() {
 				go func() {
 					// return pre-consumed quota
 					// return pre-consumed quota
 					err := model.PostConsumeTokenQuota(tokenId, -preConsumedQuota)
 					err := model.PostConsumeTokenQuota(tokenId, -preConsumedQuota)
@@ -354,6 +355,7 @@ func relayTextHelper(c *gin.Context, relayMode int) *OpenAIErrorWithStatusCode {
 						common.SysError("error return pre-consumed quota: " + err.Error())
 						common.SysError("error return pre-consumed quota: " + err.Error())
 					}
 					}
 				}()
 				}()
+			}
 			return relayErrorHandler(resp)
 			return relayErrorHandler(resp)
 		}
 		}
 	}
 	}