Quick Start

CoffeeRouter is a unified AI model gateway, fully compatible with the OpenAI API format. With a single API Key and Base URL, you can access leading AI models such as GPT, Claude, Gemini, DeepSeek, and more without managing separate accounts or credentials for each provider.

Whether you call the API directly from code or use client tools like Claude Code or OpenClaw, the setup is always the same.

Part 1 — Get Up and Running in 5 Minutes

Step 1 — Get Your API Key

1.1 Log In to the CoffeeRouter Console

Open https://coffeerouter.ai and sign in with your Google, GitHub, or Discord account.

1.2 Go to API Key Manager and Create an API Key

Click API Key Manager in the left sidebar.

Click the Create API Key button. In the dialog, configure:

• Key Name: give this key a descriptive name, such as my-first-key
• Group: optional, organize keys by project or purpose
• Quantity: number of keys to create, default is 1
• Quota Settings: choose unlimited quota or set a per-request limit
• Other Options: configure expiry date, IP whitelist, or other restrictions as needed

Click Submit and your API Key will be generated immediately.

1.3 Copy Your API Key

Once created, find your API Key in the list. Right-click on it to see your options:

• Copy Key: copy the API Key, which starts with sk-
• Copy Link Info: one-click copy of the complete configuration including Base URL and API Key

You will now have:

• Base URL: https://coffeerouter.ai, your API endpoint for all calls
• API Key: starts with sk-

Important

Your API Key is displayed in a popup right after creation. We recommend copying and saving it to a secure location immediately. You can click Show Key in the token list to view it again later.

---

Step 2 — Make Your First API Call

CoffeeRouter is fully compatible with the OpenAI API format. By modifying the configuration, you can use the standard OpenAI SDK, or any third-party clients and tools compatible with OpenAI, to seamlessly call all leading AI models aggregated by CoffeeRouter.

2.1 Browse Available Models

You can browse all available models on the Model Square.

If you need to query the available models programmatically, call the models endpoint directly:

curl https://coffeerouter.ai/v1/models \
  -H "Authorization: Bearer sk-YOUR_API_KEY"

The id field in each returned object is the model name to use.

2.2 Make Your First Call

Pick a model and fill in the model parameter to initiate a call. Here is an example using gemini-2.5-flash:

Tip

Set the model parameter to the model name you want, such as gemini-2.5-flash. Browse all models on the Model Square or query GET /v1/models.

curl https://coffeerouter.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-YOUR_API_KEY" \
  -d '{
    "model": "gemini-2.5-flash",
    "messages": [
      {"role": "user", "content": "Hello! Introduce yourself."}
    ]
  }'

from openai import OpenAI

client = OpenAI(
    base_url="https://coffeerouter.ai/v1",
    api_key="sk-YOUR_API_KEY"
)

response = client.chat.completions.create(
    model="gemini-2.5-flash",
    messages=[
        {"role": "user", "content": "Hello! Introduce yourself."}
    ]
)

print(response.choices[0].message.content)

---

Step 3 — Connect Your Tools

CoffeeRouter works with any OpenAI-compatible client or tool. You need three things:

Parameter	Value
API Address (Base URL)	https://coffeerouter.ai
API Key	The token you created in the console
Model Name	From /v1/models, or the console model list

Claude Code / Codex CLI

Command-line code assistants.

When using Claude Code or Codex CLI in your terminal, set the following environment variables:

# Claude Code
export ANTHROPIC_BASE_URL="https://coffeerouter.ai"
export ANTHROPIC_API_KEY="sk-YOUR_API_KEY"

# Codex CLI
export OPENAI_BASE_URL="https://coffeerouter.ai/v1"
export OPENAI_API_KEY="sk-YOUR_API_KEY"

Full guide: Claude Code · Codex CLI

OpenClaw

Self-hosted AI assistant platform, recommended for advanced users.

OpenClaw is a self-hosted AI assistant platform with multi-channel support. Add CoffeeRouter as a model provider in ~/.openclaw/openclaw.json:

{
  "models": {
    "mode": "merge",
    "providers": {
      "coffeerouter": {
        "baseUrl": "https://coffeerouter.ai/v1",
        "apiKey": "sk-YOUR_API_KEY",
        "api": "openai-completions",
        "models": [
          { "id": "gemini-2.5-flash", "name": "Gemini 2.5 Flash" },
          { "id": "claude-sonnet-4-6", "name": "Claude Sonnet 4.6" }
        ]
      }
    }
  },
  "agents": {
    "defaults": {
      "model": {
        "primary": "coffeerouter/gemini-2.5-flash"
      }
    }
  }
}

Full guide: OpenClaw

Cherry Studio

Desktop AI client, recommended for beginners.

Cherry Studio is a feature-rich desktop AI chat client that supports multi-model conversations.

1. Download and install Cherry Studio: https://cherry-ai.com/download
2. Open Settings → Model Providers → Add Provider
3. Set provider type to OpenAI or OpenAI-compatible
4. API Address: https://coffeerouter.ai
5. API Key: your CoffeeRouter token
6. Add the model IDs you want to use, query /v1/models for the full list

One-click Auto-fill

The CoffeeRouter console Token Manager supports one-click auto-fill for Cherry Studio. Click the button on the token list page and Cherry Studio will be configured automatically.

Full guide: Cherry Studio

Any Other OpenAI-Compatible Tool

For any tool that allows a custom API endpoint, configure:

• API Address / Base URL → https://coffeerouter.ai
• API Key → your CoffeeRouter token
• Model Name → from /v1/models

Browse all verified app integration guides: Connect Agent Tools

---

Part 2 — API Capabilities Overview

CoffeeRouter exposes the following AI model APIs, all in OpenAI-compatible format:

API Endpoint	Description
Chat Completions POST /v1/chat/completions	Multi-turn conversations, streaming, Tool Calling, structured outputs
Text Completions POST /v1/completions	Classic single-turn text completion
Image Generation POST /v1/images/generations	AI image generation
Video Generation POST /v1/videos	AI video generation
Model List GET /v1/models	Query all currently available models

Full API reference: API Docs

---

Part 3 — FAQ