Tokelix API
接入指南

Tokelix 提供 OpenAI 与 Anthropic 兼容 API。将官方 SDK 的 Base URL 指向 Tokelix，即可通过统一网关连接 GPT、Claude 和 DeepSeek。可用模型、价格与协议能力以已配置渠道为准。

Base URL: https://router2.mvu.app
身份认证: API Key (sk-…)
协议: Anthropic (/v1/messages), OpenAI (/v1/chat/completions), Responses (/v1/responses)

使用你自己的 API Key。请勿提交到代码仓库，或放入网页前端。协议能力以已配置渠道为准。

1. 身份认证

根据使用的模型选择兼容协议：

https://router2.mvu.app/v1

不同协议需要不同的身份认证请求头：

协议	请求头
Anthropic	`x-api-key: $TOKELIX_API_KEY` `anthropic-version: 2023-06-01`
OpenAI	`Authorization: Bearer $TOKELIX_API_KEY`

2. 可用模型

通过 GET /v1/models 实时获取模型列表：

BASH

curl https://router2.mvu.app/v1/models \
  -H "Authorization: Bearer $TOKELIX_API_KEY"

厂商	Model ID
OpenAI	`gpt-5-6-luna` `gpt-5-6-terra` `gpt-5-2` `gpt-5-5` `gpt-5-6-sol` `gpt-6-astra`
DeepSeek	`deepseek-v4-1-flash`
Google	`gemini-3-8-flash-openai` `gemini-3-7-flash-openai` `gemini-3-6-flash-openai` `gemini-3-5-flash-openai` `gemini-3-1-pro`
Anthropic	`claude-haiku-4-5` `claude-sonnet-4-6` `claude-sonnet-5` `claude-opus-4-7` `claude-opus-4-6` `claude-opus-4-5` `claude-fable-5` `claude-opus-5` `claude-opus-4-8`

3. Anthropic 原生格式

Endpoint: POST /v1/messages

3.1 基础请求

BASH

curl https://router2.mvu.app/v1/messages \
  -H "x-api-key: $TOKELIX_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "claude-haiku-4-5",
  "messages": [
    {
      "role": "user",
      "content": "Hello"
    }
  ],
  "max_tokens": 1024
}'

响应结构示例（占位内容）：

JSON

{
  "id": "msg_example",
  "type": "message",
  "role": "assistant",
  "model": "claude-haiku-4-5",
  "content": [
    {
      "type": "text",
      "text": "Hello!"
    }
  ],
  "stop_reason": "end_turn",
  "usage": {
    "input_tokens": 12,
    "output_tokens": 3
  }
}

3.2 SSE 流式输出

添加 stream: true，响应格式为 text/event-stream。

BASH

curl https://router2.mvu.app/v1/messages \
  -H "x-api-key: $TOKELIX_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "claude-haiku-4-5",
  "messages": [
    {
      "role": "user",
      "content": "Hello"
    }
  ],
  "max_tokens": 1024,
  "stream": true
}'

3.3 Python SDK

PYTHON

from anthropic import Anthropic
import os

client = Anthropic(
    api_key=os.environ["TOKELIX_API_KEY"],
    base_url="https://router2.mvu.app",
)
response = client.messages.create(
    model="claude-haiku-4-5",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.content[0].text)

4. OpenAI 兼容格式

Endpoints: POST /v1/chat/completions · POST /v1/responses

4.1 Chat Completions

BASH

curl https://router2.mvu.app/v1/chat/completions \
  -H "Authorization: Bearer $TOKELIX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "gpt-5-6-luna",
  "messages": [
    {
      "role": "user",
      "content": "Hello"
    }
  ]
}'

响应结构示例（占位内容）：

JSON

{
  "object": "chat.completion",
  "model": "gpt-5-6-luna",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Hello!"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 12,
    "completion_tokens": 3,
    "total_tokens": 15
  }
}

4.2 Python SDK

PYTHON

from openai import OpenAI

client = OpenAI(
    base_url="https://router2.mvu.app/v1",
    api_key="<TOKELIX_API_KEY>",
)
response = client.chat.completions.create(
    model="gpt-5-6-luna",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

4.3 流式输出

BASH

curl https://router2.mvu.app/v1/chat/completions \
  -H "Authorization: Bearer $TOKELIX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "gpt-5-6-luna",
  "messages": [
    {
      "role": "user",
      "content": "Hello"
    }
  ],
  "stream": true
}'

Chat Completions 使用 data: {...} SSE 消息，并以 data: [DONE] 结束。

4.4 Responses API

BASH

curl https://router2.mvu.app/v1/responses \
  -H "Authorization: Bearer $TOKELIX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "gpt-5-6-luna",
  "input": "Hello"
}'

4.5 Responses 流式输出

BASH

curl https://router2.mvu.app/v1/responses \
  -H "Authorization: Bearer $TOKELIX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "gpt-5-6-luna",
  "input": "Hello",
  "stream": true
}'

5. 接入 AI 编程工具

Claude Code 使用 Anthropic 协议；Codex 使用 Responses 协议。

BASH

export ANTHROPIC_BASE_URL="https://router2.mvu.app"
export ANTHROPIC_AUTH_TOKEN="$TOKELIX_API_KEY"
export ANTHROPIC_MODEL="claude-haiku-4-5"
claude

Codex CLI · Claude Code · DeepSeek

6. FAQ

认证失败？

核对请求头与 Base URL，确认 Key 未过期或被删除，且没有额外空格。OpenAI SDK 的地址需要 /v1，Anthropic SDK 不需要。

模型不可用？

先调用 /v1/models 核对模型 ID，再检查组织的路由限制、余额和预算。

响应较慢？

开启 stream: true，设置足够的客户端读取超时。失败时记录请求 ID，避免无上限重复请求。

准备好了？三步开始

登录控制台 · 联系管理员充值 · 创建 API Key

获取 API Key

下一篇模型与路由