Free LLM Gateway — AI skill for Claude Code
Local LiteLLM gateway aggregating free LLM providers (Zhipu/SiliconFlow/OpenRouter/NVIDIA/Agnes/SenseNova/dots) with automatic 429 failover — one OpenAI-compatible endpoint.
How to install Free LLM Gateway
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open heavenchenggong/free-llm-gateway and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Free LLM Gateway does
Local LiteLLM gateway aggregating free LLM providers (Zhipu/SiliconFlow/OpenRouter/NVIDIA/Agnes/SenseNova/dots) with automatic 429 failover — one OpenAI-compatible endpoint.
Alternatives in AI
- Axonhub — ⚡️ Open-source AI Gateway — Use any SDK to call 100+ LLMs 3.2k ★
- Cc Mirror — Create multiple isolated Claude Code variants with custom providers (Z.ai, MiniMax, OpenRouter, LiteLLM) 2.1k ★
- My Free Code — Open-source multi-provider AI gateway for Claude Code and other coding agents, with model routing, streaming 633 ★
README
free-llm-gateway
本地 LiteLLM 网关,把多个免费 LLM API 渠道聚合成一个 OpenAI 兼容端点——任一渠道 限流(429)、超时或报错时自动冷却 30 秒并切换下一个,客户端只配一个模型。
[English] A local LiteLLM gateway that aggregates multiple free LLM API providers (Zhipu BigModel, SiliconFlow, OpenRouter, NVIDIA NIM, Agnes AI, SenseNova, dots.ai) behind one OpenAI-compatible endpoint with automatic failover on 429/timeout.
它解决什么问题
免费渠道各自单独配在客户端里,一个限流就要手动换模型。装了这个网关:
客户端(只配一个模型)
→ 本地网关 http://127.0.0.1:4010/v1
→ 渠道A / 渠道B / 渠道C ...(失败自动冷却并切换)
快速开始
# 1. 拿 key(2-3 个渠道即可起步,注册直达链接见 references/channels.md)
# 2. 一键安装(macOS,Terminal 里跑)
bash scripts/setup_gateway.sh
# 3. 验证
bash scripts/verify.sh
# 4. 客户端配置(WorkBuddy / Claude Code / 任意 OpenAI 兼容工具)
# 接口地址: http://127.0.0.1:4010/v1
# API Key: free(任意非空)
# 模型 ID: free-chat / free-vision / free-embed
安装为 WorkBuddy / Claude Code skill
把整个目录放进 `~/.workbuddy/skills/`(WorkBuddy)或 `~/.claude/skills/`(Claude Code), 对话里说"帮我装免费模型网关"即可触发。
文档
| 文件 | 内容 |
|---|---|
SKILL.md |
安装/配置主流程 + 自动化边界 |
references/channels.md |
7 家免费渠道注册直达链接、模型 ID、每家的坑 |
references/troubleshooting.md |
安装/常驻/运行三阶段踩坑全集 + 状态码速查 |
已知限制
- launchd 常驻注册需在用户自己的 Terminal 执行(Agent/沙箱会话会被系统拒绝,脚本会自动降级并提示)
- 免费渠道高峰期可能多个同时 429,重试耗尽后单次请求仍可能失败——重发即可
- 各平台免费政策变动频繁(2026-09 口径),接入前先用真实 chat 请求验证 key
License
MIT
Related Skills
Freeroute
Get up to 50M+ or virtually unlimited free AI tokens daily! Universal OpenAI gateway that pools free-tier mode
SRouter
Local-first AI Gateway: Connect Claude Code, Cursor, Windsurf, & Cline to Antigravity, OpenAI, Claude, & Gemin
Anthropic Proxy Rs
A proxy server that intercepts Anthropic API requests and converts them to OpenAI-compatible format, enabling
Stone
Personal, local-first desktop AI gateway for Windows, macOS, and Linux. Connect Claude Code, Codex, and Gemini
Agent Observability Stack
Self-hostable observability for LLM agents + a Linux host (Intel iGPU/NPU + NVIDIA RTX 3090 eGPU aware): Prome
Clipal
Minimal LLM API reverse proxy for Claude Code, Codex CLI & Gemini CLI — YAML routing, hot reload, and automati
Related Agents
Openrouter Agent
Runs a model from any of ~60 vendors through OpenRouter's OpenAI-compatible endpoint, billed per token against
LLM Orchestrator
Use this agent for LLM integration work — prompt engineering, multi-provider abstraction (OpenRouter, Gemini,
Codex CLI
Codex CLI → any OpenAI-compatible endpoint