phsl / new-api

Continue fixing Ollama embedding return issue

Modify ollama embed return fields

(cherry picked from commit 90daa38d5bea7b158ebed9990f042f6bf8567eb3)

(cherry picked from commit ec79110c99e9b4c076c5f7b8285e535b9c5052db)

feat: support openai realtime api

(cherry picked from commit fdfea8726c6d86d3844af1ac18d7b3df908f26a7)

(cherry picked from commit 052bdab1c45b3a4ba5f079afc763f54e751b1cd7)

(cherry picked from commit 618dffc43fd5a5f4065944db87761f9ee18e44d3)

(cherry picked from commit fd24dc467bfc360008b313220e607f0176ee7aa3)

(cherry picked from commit 96373455521a38095706bd81c57f9a18557d9c2e)

(cherry picked from commit 273d154e1640bae26b7caedddf1685e9ff21ab74)

(cherry picked from commit d87917f8f6eb9d2e144a9f840d6d91767ea2eb69)

(cherry picked from commit 75ff3d98f06103dc2df1f8817bd3fcbf433e0f20)

(cherry picked from commit d4966246e68dbdcdab45ec5c5141362834d74425)

(cherry picked from commit a5529df3e1a4c08a120e8c05203a7d885b0fe8d8)

(cherry picked from commit 030187ff75c64c40017cda2fa98ef2b3c01f0bd5)

chore: 更新最新haiku模型倍率

ollama /api/embeddings is deprecated, use /api/embed.

/api/embeddings is deprecated, use /api/embed.

chore: 修正chatgpt-4o-latest补全倍率

fix wrong model ration about chatgpt-4o-latest

(cherry picked from commit 8d41c17ccf19cb29100dbe506d3d42a6be822ff9)