接口参考

接口参考。

先看端点，再看参数，再看响应。

打开在线调试返回总览页

API 文档/接口参考

端点

先按供应商找入口。

四家供应商分路由接入。

Base URL

https://tokenapi.bytrabbit.com

GET /v1/models

只返回当前账号可访问的 OpenAI 模型。

POST /v1/responses

OpenAI Responses 原生入口，常用子资源也已补齐。

POST /v1/chat/completions

只服务 OpenAI 原生聊天请求。

POST /v1/embeddings

OpenAI 文本向量嵌入入口。

GET/POST /v1/fine_tuning/jobs

OpenAI 微调任务入口，支持列表、创建、详情、取消和事件查询。

POST /v1/moderations

OpenAI 内容审核入口。

POST /v1/realtime/sessions

创建 Realtime session。HTTP baseUrl 不变；若已部署 relay，响应头会返回 `x-tokenai-realtime-relay-url` 供客户端自动发现 WebSocket 地址。

POST /v1/messages

Claude 原生入口；`messages`、`count_tokens`、`batches`、`files` 已支持。

POST /v1beta/models/{model}:generateContent

Gemini generateContent 原生入口。

POST /v1beta/models/{model}:streamGenerateContent

Gemini 原生流式生成入口。

GET /compatible-mode/v1/models

Qwen DashScope compatible-mode 模型列表入口。

POST /compatible-mode/v1/chat/completions

Qwen DashScope compatible-mode 聊天入口。

POST /compatible-mode/v1/embeddings

Qwen DashScope compatible-mode 向量嵌入入口。

POST /compatible-mode/v1/files

Qwen DashScope compatible-mode 文件入口，配合 batches 使用。

POST /compatible-mode/v1/batches

Qwen DashScope compatible-mode Batch 入口。

POST /compatible-mode/v1/responses

Qwen DashScope compatible-mode Responses 入口。

正式请求地址

Base URL
https://api.ciyuan.com

OpenAI
GET /v1/models
POST /v1/chat/completions
POST /v1/responses
POST /v1/embeddings
POST /v1/moderations
GET|POST /v1/fine_tuning/jobs
POST /v1/realtime/sessions
GET|POST /v1/assistants
POST /v1/threads
GET|POST /v1/files
POST /v1/uploads
GET|POST /v1/batches
GET|POST /v1/vector_stores
POST /v1/images/generations
POST /v1/audio/speech

Anthropic
GET /v1/models
POST /v1/messages
POST /v1/messages/count_tokens
GET|POST /v1/messages/batches
GET|POST /v1/files

Gemini
GET /v1beta/models/{model}
POST /v1beta/models/{model}:generateContent
POST /v1beta/models/{model}:streamGenerateContent
POST /v1beta/models/{model}:countTokens
POST /v1beta/models/{model}:embedContent
GET|POST /v1beta/files
POST /upload/v1beta/files
GET|POST /v1beta/cachedContents
GET /v1beta/batches
GET|POST /v1beta/tunedModels

Qwen compatible-mode
GET /compatible-mode/v1/models
POST /compatible-mode/v1/chat/completions
POST /compatible-mode/v1/embeddings
GET|POST /compatible-mode/v1/files
GET|POST /compatible-mode/v1/batches
POST /compatible-mode/v1/responses
POST /compatible-mode/v1/conversations

Qwen native
POST /api/v1/services/aigc/text-generation/generation
POST /api/v1/services/embeddings/text-embedding/text-embedding
POST /api/v1/services/aigc/multimodal-generation/generation
POST /api/v1/services/aigc/text2image/image-synthesis
GET|POST /api/v1/tasks/{task_id}

参数

这里只列稳定约束。

参数	要求	说明
OpenAI /v1/*	原生	多数路由是 OpenAI 原生字段；`/v1/models` 与 `/v1/files` 会按绑定供应商分流。
Anthropic /v1/messages*	原生	使用 Claude Messages 原生字段。
Shared /v1/models* /v1/files*	按供应商分流	Anthropic 绑定 token 会走 Claude Models / Files；OpenAI 绑定 token 走 OpenAI 原生接口。
Gemini /v1beta/*	原生	使用 Gemini `contents` 与 `parts` 原生字段。
Qwen /compatible-mode/v1/*	原生	使用 DashScope compatible-mode 原生字段。
Qwen /api/v1/services/aigc/text-generation/generation	原生	使用 DashScope 原生 Generation 字段。
Authorization	必填	所有供应商都使用 `Bearer tk_xxx`。

响应

返回结构也要看。

OpenAI /v1/models

{
  "object": "list",
  "data": [
    { "id": "byt-gpt-5.4", "object": "model" },
    { "id": "byt-gpt-4.1-mini", "object": "model" }
  ]
}

OpenAI chat response

{
  "id": "chatcmpl_xxx",
  "object": "chat.completion",
  "model": "byt-gpt-5.4",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "你好，这是一条 OpenAI 原生返回结果。"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 18,
    "completion_tokens": 28,
    "total_tokens": 46
  }
}

stream=true 的 SSE chunk

data: {
  "id": "chatcmpl_xxx",
  "object": "chat.completion.chunk",
  "choices": [
    {
      "index": 0,
      "delta": { "content": "你好" },
      "finish_reason": null
    }
  ]
}

data: [DONE]

Anthropic message batch create

POST https://api.ciyuan.com/v1/messages/batches
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "requests": [
    {
      "custom_id": "job-1",
      "params": {
        "model": "byt-claude-sonnet-4-20250514",
        "max_tokens": 128,
        "messages": [
          { "role": "user", "content": "Summarize this ticket." }
        ]
      }
    }
  ]
}

Anthropic files upload

POST https://api.ciyuan.com/v1/files
Authorization: Bearer tk_xxx
anthropic-version: 2023-06-01
anthropic-beta: files-api-2025-04-14
Content-Type: multipart/form-data

purpose=user_data
[email protected]

Gemini embedContent

POST https://api.ciyuan.com/v1beta/models/byt-text-embedding-004:embedContent
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "content": {
    "parts": [
      { "text": "CIYUAN AI lets teams route requests across providers." }
    ]
  }
}

Gemini batchEmbedContents

POST https://api.ciyuan.com/v1beta/models/byt-text-embedding-004:batchEmbedContents
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "requests": [
    {
      "content": {
        "parts": [
          { "text": "First sentence." }
        ]
      }
    },
    {
      "content": {
        "parts": [
          { "text": "Second sentence." }
        ]
      }
    }
  ]
}

Gemini file metadata create

POST https://api.ciyuan.com/v1beta/files
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "file": {
    "displayName": "notes.txt"
  }
}

Gemini files upload start

POST https://api.ciyuan.com/upload/v1beta/files
Authorization: Bearer tk_xxx
X-Goog-Upload-Protocol: resumable
X-Goog-Upload-Command: start
X-Goog-Upload-Header-Content-Length: 1234
X-Goog-Upload-Header-Content-Type: text/plain
Content-Type: application/json

{
  "file": {
    "display_name": "notes.txt"
  }
}

Gemini cached content create

POST https://api.ciyuan.com/v1beta/cachedContents
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "model": "models/byt-gemini-2.0-flash-001",
  "displayName": "support-doc-cache",
  "contents": [
    {
      "role": "user",
      "parts": [
        { "text": "Long support policy text..." }
      ]
    }
  ],
  "ttl": "3600s"
}

Gemini batch generate content

POST https://api.ciyuan.com/v1beta/models/byt-gemini-2.0-flash:batchGenerateContent
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "batch": {
    "displayName": "support-eval-batch",
    "inputConfig": {
      "requests": [
        {
          "contents": [
            {
              "role": "user",
              "parts": [{ "text": "Summarize ticket A" }]
            }
          ]
        }
      ]
    }
  }
}

Gemini tuned models list

GET https://api.ciyuan.com/v1beta/tunedModels?pageSize=20
Authorization: Bearer tk_xxx

Qwen native generation

POST https://api.ciyuan.com/api/v1/services/aigc/text-generation/generation
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "model": "byt-qwen-plus",
  "input": {
    "messages": [
      { "role": "user", "content": "Write a short product summary." }
    ]
  },
  "parameters": {
    "result_format": "message"
  }
}

Qwen native embedding

POST https://api.ciyuan.com/api/v1/services/embeddings/text-embedding/text-embedding
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "model": "byt-text-embedding-v3",
  "input": {
    "texts": [
      "CIYUAN AI routes requests across providers."
    ]
  }
}

Qwen native multimodal

POST https://api.ciyuan.com/api/v1/services/aigc/multimodal-generation/generation
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "model": "byt-qwen-vl-max",
  "input": {
    "messages": [
      {
        "role": "user",
        "content": [
          { "image": "https://example.com/demo.png" },
          { "text": "Describe this image in one paragraph." }
        ]
      }
    ]
  }
}

Qwen native image async

POST https://api.ciyuan.com/api/v1/services/aigc/text2image/image-synthesis
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "model": "byt-wanx2.1-t2i-turbo",
  "input": {
    "prompt": "A cinematic night city street with rain reflections."
  }
}

GET https://api.ciyuan.com/api/v1/tasks/{task_id}

Qwen task cancel

POST https://api.ciyuan.com/api/v1/tasks/{task_id}/cancel
Authorization: Bearer tk_xxx
Content-Type: application/json

{}

路由矩阵

各供应商各走原生入口。

能力	状态	说明
GET /v1/models* / /v1/files*	OpenAI / Anthropic	共享入口，按 token 绑定供应商分流。
OpenAI /v1/chat/completions / /v1/responses*	OpenAI	文本主路径。
OpenAI /v1/embeddings / /v1/moderations	OpenAI	向量与审核。
OpenAI /v1/assistants* / /v1/threads*	OpenAI	Assistants、Threads、Runs、steps、tool outputs。
OpenAI /v1/uploads* / /v1/batches*	OpenAI	上传与批处理。
OpenAI /v1/vector_stores*	OpenAI	向量库、文件挂载、批次、搜索。
OpenAI /v1/images/* / /v1/audio/*	OpenAI	图片与音频主路径。
OpenAI /v1/fine_tuning/jobs* / /v1/realtime/sessions	OpenAI	微调任务与 Realtime session。
Anthropic /v1/messages / /v1/messages/count_tokens	Anthropic	Claude 消息与 token 统计。
Anthropic /v1/messages/batches*	Anthropic	批量消息：列表、创建、详情、取消、results。
Gemini /v1beta/models/*:generateContent / :streamGenerateContent	Gemini	生成主路径。
Gemini /v1beta/models/*:countTokens / :embedContent / :batchEmbedContents	Gemini	token 统计与嵌入。
Gemini /v1beta/files* / /upload/v1beta/files	Gemini	文件列表、metadata、resumable 上传。
Gemini /v1beta/cachedContents* / /v1beta/batches*	Gemini	缓存与批处理。
Gemini /v1beta/tunedModels* / Gemini Live relay	Gemini	tunedModels 和 Live relay。
Qwen /compatible-mode/v1/models / chat/completions / embeddings	Qwen	compatible-mode 主路径。
Qwen /compatible-mode/v1/files* / batches* / responses / conversations*	Qwen	compatible-mode 文件、批处理、Responses、会话。
Qwen /api/v1/services/aigc/* / embeddings/* / tasks/*	Qwen	DashScope 原生 generation、multimodal、image async、embedding、task。

接口参考

接口参考。

先看端点，再看参数，再看响应。

打开在线调试返回总览页

API 文档/接口参考

端点

先按供应商找入口。

四家供应商分路由接入。

Base URL

https://tokenapi.bytrabbit.com

GET /v1/models

只返回当前账号可访问的 OpenAI 模型。

POST /v1/responses

OpenAI Responses 原生入口，常用子资源也已补齐。

POST /v1/chat/completions

只服务 OpenAI 原生聊天请求。

POST /v1/embeddings

OpenAI 文本向量嵌入入口。

GET/POST /v1/fine_tuning/jobs

OpenAI 微调任务入口，支持列表、创建、详情、取消和事件查询。

POST /v1/moderations

OpenAI 内容审核入口。

POST /v1/realtime/sessions

创建 Realtime session。HTTP baseUrl 不变；若已部署 relay，响应头会返回 `x-tokenai-realtime-relay-url` 供客户端自动发现 WebSocket 地址。

POST /v1/messages

Claude 原生入口；`messages`、`count_tokens`、`batches`、`files` 已支持。

POST /v1beta/models/{model}:generateContent

Gemini generateContent 原生入口。

POST /v1beta/models/{model}:streamGenerateContent

Gemini 原生流式生成入口。

GET /compatible-mode/v1/models

Qwen DashScope compatible-mode 模型列表入口。

POST /compatible-mode/v1/chat/completions

Qwen DashScope compatible-mode 聊天入口。

POST /compatible-mode/v1/embeddings

Qwen DashScope compatible-mode 向量嵌入入口。

POST /compatible-mode/v1/files

Qwen DashScope compatible-mode 文件入口，配合 batches 使用。

POST /compatible-mode/v1/batches

Qwen DashScope compatible-mode Batch 入口。

POST /compatible-mode/v1/responses

Qwen DashScope compatible-mode Responses 入口。

正式请求地址

Base URL
https://api.ciyuan.com

OpenAI
GET /v1/models
POST /v1/chat/completions
POST /v1/responses
POST /v1/embeddings
POST /v1/moderations
GET|POST /v1/fine_tuning/jobs
POST /v1/realtime/sessions
GET|POST /v1/assistants
POST /v1/threads
GET|POST /v1/files
POST /v1/uploads
GET|POST /v1/batches
GET|POST /v1/vector_stores
POST /v1/images/generations
POST /v1/audio/speech

Anthropic
GET /v1/models
POST /v1/messages
POST /v1/messages/count_tokens
GET|POST /v1/messages/batches
GET|POST /v1/files

Gemini
GET /v1beta/models/{model}
POST /v1beta/models/{model}:generateContent
POST /v1beta/models/{model}:streamGenerateContent
POST /v1beta/models/{model}:countTokens
POST /v1beta/models/{model}:embedContent
GET|POST /v1beta/files
POST /upload/v1beta/files
GET|POST /v1beta/cachedContents
GET /v1beta/batches
GET|POST /v1beta/tunedModels

Qwen compatible-mode
GET /compatible-mode/v1/models
POST /compatible-mode/v1/chat/completions
POST /compatible-mode/v1/embeddings
GET|POST /compatible-mode/v1/files
GET|POST /compatible-mode/v1/batches
POST /compatible-mode/v1/responses
POST /compatible-mode/v1/conversations

Qwen native
POST /api/v1/services/aigc/text-generation/generation
POST /api/v1/services/embeddings/text-embedding/text-embedding
POST /api/v1/services/aigc/multimodal-generation/generation
POST /api/v1/services/aigc/text2image/image-synthesis
GET|POST /api/v1/tasks/{task_id}

参数

这里只列稳定约束。

参数	要求	说明
OpenAI /v1/*	原生	多数路由是 OpenAI 原生字段；`/v1/models` 与 `/v1/files` 会按绑定供应商分流。
Anthropic /v1/messages*	原生	使用 Claude Messages 原生字段。
Shared /v1/models* /v1/files*	按供应商分流	Anthropic 绑定 token 会走 Claude Models / Files；OpenAI 绑定 token 走 OpenAI 原生接口。
Gemini /v1beta/*	原生	使用 Gemini `contents` 与 `parts` 原生字段。
Qwen /compatible-mode/v1/*	原生	使用 DashScope compatible-mode 原生字段。
Qwen /api/v1/services/aigc/text-generation/generation	原生	使用 DashScope 原生 Generation 字段。
Authorization	必填	所有供应商都使用 `Bearer tk_xxx`。

响应

返回结构也要看。

OpenAI /v1/models

{
  "object": "list",
  "data": [
    { "id": "byt-gpt-5.4", "object": "model" },
    { "id": "byt-gpt-4.1-mini", "object": "model" }
  ]
}

OpenAI chat response

{
  "id": "chatcmpl_xxx",
  "object": "chat.completion",
  "model": "byt-gpt-5.4",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "你好，这是一条 OpenAI 原生返回结果。"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 18,
    "completion_tokens": 28,
    "total_tokens": 46
  }
}

stream=true 的 SSE chunk

data: {
  "id": "chatcmpl_xxx",
  "object": "chat.completion.chunk",
  "choices": [
    {
      "index": 0,
      "delta": { "content": "你好" },
      "finish_reason": null
    }
  ]
}

data: [DONE]

Anthropic message batch create

POST https://api.ciyuan.com/v1/messages/batches
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "requests": [
    {
      "custom_id": "job-1",
      "params": {
        "model": "byt-claude-sonnet-4-20250514",
        "max_tokens": 128,
        "messages": [
          { "role": "user", "content": "Summarize this ticket." }
        ]
      }
    }
  ]
}

Anthropic files upload

POST https://api.ciyuan.com/v1/files
Authorization: Bearer tk_xxx
anthropic-version: 2023-06-01
anthropic-beta: files-api-2025-04-14
Content-Type: multipart/form-data

purpose=user_data
[email protected]

Gemini embedContent

POST https://api.ciyuan.com/v1beta/models/byt-text-embedding-004:embedContent
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "content": {
    "parts": [
      { "text": "CIYUAN AI lets teams route requests across providers." }
    ]
  }
}

Gemini batchEmbedContents

POST https://api.ciyuan.com/v1beta/models/byt-text-embedding-004:batchEmbedContents
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "requests": [
    {
      "content": {
        "parts": [
          { "text": "First sentence." }
        ]
      }
    },
    {
      "content": {
        "parts": [
          { "text": "Second sentence." }
        ]
      }
    }
  ]
}

Gemini file metadata create

POST https://api.ciyuan.com/v1beta/files
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "file": {
    "displayName": "notes.txt"
  }
}

Gemini files upload start

POST https://api.ciyuan.com/upload/v1beta/files
Authorization: Bearer tk_xxx
X-Goog-Upload-Protocol: resumable
X-Goog-Upload-Command: start
X-Goog-Upload-Header-Content-Length: 1234
X-Goog-Upload-Header-Content-Type: text/plain
Content-Type: application/json

{
  "file": {
    "display_name": "notes.txt"
  }
}

Gemini cached content create

POST https://api.ciyuan.com/v1beta/cachedContents
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "model": "models/byt-gemini-2.0-flash-001",
  "displayName": "support-doc-cache",
  "contents": [
    {
      "role": "user",
      "parts": [
        { "text": "Long support policy text..." }
      ]
    }
  ],
  "ttl": "3600s"
}

Gemini batch generate content

POST https://api.ciyuan.com/v1beta/models/byt-gemini-2.0-flash:batchGenerateContent
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "batch": {
    "displayName": "support-eval-batch",
    "inputConfig": {
      "requests": [
        {
          "contents": [
            {
              "role": "user",
              "parts": [{ "text": "Summarize ticket A" }]
            }
          ]
        }
      ]
    }
  }
}

Gemini tuned models list

GET https://api.ciyuan.com/v1beta/tunedModels?pageSize=20
Authorization: Bearer tk_xxx

Qwen native generation

POST https://api.ciyuan.com/api/v1/services/aigc/text-generation/generation
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "model": "byt-qwen-plus",
  "input": {
    "messages": [
      { "role": "user", "content": "Write a short product summary." }
    ]
  },
  "parameters": {
    "result_format": "message"
  }
}

Qwen native embedding

POST https://api.ciyuan.com/api/v1/services/embeddings/text-embedding/text-embedding
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "model": "byt-text-embedding-v3",
  "input": {
    "texts": [
      "CIYUAN AI routes requests across providers."
    ]
  }
}

Qwen native multimodal

POST https://api.ciyuan.com/api/v1/services/aigc/multimodal-generation/generation
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "model": "byt-qwen-vl-max",
  "input": {
    "messages": [
      {
        "role": "user",
        "content": [
          { "image": "https://example.com/demo.png" },
          { "text": "Describe this image in one paragraph." }
        ]
      }
    ]
  }
}

Qwen native image async

POST https://api.ciyuan.com/api/v1/services/aigc/text2image/image-synthesis
Authorization: Bearer tk_xxx
Content-Type: application/json

{
  "model": "byt-wanx2.1-t2i-turbo",
  "input": {
    "prompt": "A cinematic night city street with rain reflections."
  }
}

GET https://api.ciyuan.com/api/v1/tasks/{task_id}

Qwen task cancel

POST https://api.ciyuan.com/api/v1/tasks/{task_id}/cancel
Authorization: Bearer tk_xxx
Content-Type: application/json

{}

路由矩阵

各供应商各走原生入口。

能力	状态	说明
GET /v1/models* / /v1/files*	OpenAI / Anthropic	共享入口，按 token 绑定供应商分流。
OpenAI /v1/chat/completions / /v1/responses*	OpenAI	文本主路径。
OpenAI /v1/embeddings / /v1/moderations	OpenAI	向量与审核。
OpenAI /v1/assistants* / /v1/threads*	OpenAI	Assistants、Threads、Runs、steps、tool outputs。
OpenAI /v1/uploads* / /v1/batches*	OpenAI	上传与批处理。
OpenAI /v1/vector_stores*	OpenAI	向量库、文件挂载、批次、搜索。
OpenAI /v1/images/* / /v1/audio/*	OpenAI	图片与音频主路径。
OpenAI /v1/fine_tuning/jobs* / /v1/realtime/sessions	OpenAI	微调任务与 Realtime session。
Anthropic /v1/messages / /v1/messages/count_tokens	Anthropic	Claude 消息与 token 统计。
Anthropic /v1/messages/batches*	Anthropic	批量消息：列表、创建、详情、取消、results。
Gemini /v1beta/models/*:generateContent / :streamGenerateContent	Gemini	生成主路径。
Gemini /v1beta/models/*:countTokens / :embedContent / :batchEmbedContents	Gemini	token 统计与嵌入。
Gemini /v1beta/files* / /upload/v1beta/files	Gemini	文件列表、metadata、resumable 上传。
Gemini /v1beta/cachedContents* / /v1beta/batches*	Gemini	缓存与批处理。
Gemini /v1beta/tunedModels* / Gemini Live relay	Gemini	tunedModels 和 Live relay。
Qwen /compatible-mode/v1/models / chat/completions / embeddings	Qwen	compatible-mode 主路径。
Qwen /compatible-mode/v1/files* / batches* / responses / conversations*	Qwen	compatible-mode 文件、批处理、Responses、会话。
Qwen /api/v1/services/aigc/* / embeddings/* / tasks/*	Qwen	DashScope 原生 generation、multimodal、image async、embedding、task。

页面加载中

接口参考。

先按供应商找入口。

这里只列稳定约束。

返回结构也要看。

各供应商各走原生入口。

接口参考。

先按供应商找入口。

这里只列稳定约束。

返回结构也要看。

各供应商各走原生入口。