API Introduction

The Chatspeed ccproxy module provides flexible and powerful API access rules for easy invocation by various clients. This section details these rules, helping you understand how to access and combine various AI models and tools via ccproxy.

Core Concepts

API access in ccproxy is primarily based on the following two core concepts:

1. Grouping (Group) You can set up independent model groups for different clients or use cases. By including the group name in the API path, you can easily isolate different clients' access capabilities to models and quickly switch models as needed.

   • Example: If your group name is gemini, to access the Claude protocol chat interface, you can use /gemini/v1/messages.

2. Tool Compatibility Mode For models that do not natively support tool calling (Function Calling), ccproxy provides a tool compatibility mode, allowing them to gain and execute tool calling capabilities. To enable tool compatibility mode, simply add compat_mode to the API entry endpoint.

   • Example: /gemini/compat_mode/v1/messages

API Access Endpoints

Below are the basic API entry points provided by ccproxy:

• /sse: MCP Proxy Entry (Server-Sent Events)
• /v1/models: OpenAI compatible interface and Claude format model list interface
• /v1/chat/completions: OpenAI compatible interface chat entry
• /v1/messages: Claude native protocol message interface
• /v1beta/models: Gemini chat model list interface
• /v1beta/models/{model}/generateContent: Gemini chat synchronous access interface
• /v1beta/models/{model}/streamGenerateContent: Gemini chat streaming access interface
• /api/tags: Ollama protocol model list interface
• /api/chat: Ollama protocol chat entry
• /api/show: Ollama protocol model details interface

Combined API Access

By combining Grouping and Tool Compatibility Mode, you can build more flexible API access paths. Please refer to the API list below for specific combinations.