AI · API Platform

LongCat API Platform

AI model proxy service for the LongCat series — compatible with OpenAI and Anthropic API formats. Build agents, coding assistants, and enterprise workflows on a fast, open, agentic-first platform.

🧠 Model: LongCat-2.0-Preview 🔓 Access: Open & available now 📦 Formats: OpenAI · Anthropic ⚡ Quota: 5M tokens/day free

LongCat-2.0-Preview is a high-performance agentic model supporting both OpenAI and Anthropic API formats — tool-calling, multi-step planning, and real-world task completion baked into the weights.

What's New

The latest release in the LongCat family — a high-performance agentic model supporting both OpenAI and Anthropic API formats. Tool-calling, multi-step planning, and real-world task completion baked into the weights.

2026-05-29: Six legacy models retired (incl. Flash Thinking). LongCat-2.0-Preview is now the primary model.
2026-03-12: LongCat-Flash-Thinking upgraded for top-tier inference; auto-routed with no code changes.
2026-01-14: LongCat-Flash-Thinking-2601 released — 560B MoE architecture, upgraded reasoning.
2025-09-22: LongCat Flash Thinking officially open-sourced.

Full ChangeLog →

API Endpoints

Base URL: https://api.longcat.chat
Auth: Authorization: Bearer YOUR_API_KEY

Method	Endpoint	Description
POST	`/openai/v1/chat/completions`	Chat completions (OpenAI format)
POST	`/anthropic/v1/messages`	Messages (Anthropic format)
GET	`/openai/v1/models`	List available models
GET	`/openai/v1/models/{model}`	Model details

Key Parameters

Parameter	Type	Required	Description
`model`	string	Yes	e.g. `LongCat-2.0-Preview`
`messages`	array	Yes	Chat messages array
`max_tokens`	integer	No	Max output (up to 128K)
`stream`	boolean	No	Enable SSE streaming
`temperature`	number	No	Sampling temperature

Full API Docs →

Coding Tools

Use LongCat with your favorite coding assistant. All tools support LongCat-2.0-Preview via OpenAI-compatible API format.

Configure ~/.claude/settings.json:

ANTHROPIC_BASE_URL = https://api.longcat.chat/anthropic
ANTHROPIC_AUTH_TOKEN = your API key
ANTHROPIC_MODEL = LongCat-2.0-Preview

Claude Code Docs →

VS Code extension. Set API Provider to "OpenAI Compatible":

Base URL: https://api.longcat.chat/openai
Model: LongCat-2.0-Preview

Kilo Code Docs →

CLI tool. Configure ~/.config/opencode/opencode.json:

Provider: LongCat (OpenAI-compatible)
Base URL: https://api.longcat.chat/openai
Model: LongCat-2.0-Preview

OpenCode Docs →

Gateway-based agent platform. Install via npm, then configure:

Base URL: https://api.longcat.chat/openai
Model: LongCat-2.0-Preview (200K context)

Run openclaw tui or openclaw dashboard.

OpenClaw Docs →

OpenAI Codex CLI. Configure ~/.codex/config.toml and ~/.codex/auth.json:

Model: LongCat-2.0-Preview
Base URL: https://api.longcat.chat/openai/v1
Wire API: responses

Codex Docs →

FAQ

Log in at longcat.chat/platform/api_keys and click "Create API Key." Keys are shown only once — store securely. Up to 100 active keys per account.

/openai/native/chat/completions uses OpenAI format (system/user/assistant roles). /anthropic/v1/messages uses Claude format (separate system parameter). Same API key works for both.

LongCat-2.0-Preview outputs up to 128K tokens. 1M context window. Free tier: 5M tokens/day (up to 120M via feedback submissions).

Yes. For OpenAI: set openai.api_base = "https://api.longcat.chat/openai/". For Anthropic: configure client = anthropic.Anthropic(base_url="https://api.longcat.chat/anthropic"). Minimal code changes needed.

Set "stream": true in the request body. Responses return as Server-Sent Events (SSE).

Too many requests too quickly. Implement exponential backoff or reduce request rate. Limits are enforced per API Key.

Full FAQ →

Getting Started

Get an API Key → Documentation Benchmarks

Browse the rest of the Dump collection.