awesome-free-llm-apis — Claude Code skill: free LLM APIs
A reference skill listing LLM providers with permanent free tiers, their rate limits, models and Anthropic-compatible setup.
npx skills add reason-machines/trending-skills --skill awesome-free-llm-apis -a claude-codeWhat awesome-free-llm-apis does
The skill hands your agent a ready list of providers that offer a permanent free tier for text inference, with no trial credits and no expiry date. Two groups are covered: companies serving their own models (Cohere, Google Gemini, Mistral AI, Zhipu AI) and hosts running open-weight ones (Cerebras, Groq, Cloudflare Workers AI, GitHub Models, Hugging Face, Kluster AI, LLM7.io, NVIDIA NIM, Ollama Cloud, OpenRouter). Each entry carries notable models, request limits and region.
The second half is working code. Python and JavaScript/TypeScript samples show how to point an OpenAI SDK client at a provider by swapping base_url and api_key. Ollama Cloud is the one endpoint that does not fit that pattern. Keys belong in environment variables such as GROQ_API_KEY or GEMINI_API_KEY, not in source files.
Read the region and limit columns before you commit. Gemini's free tier does not cover the EU, UK or Switzerland. Zhipu AI and Kluster AI publish no rate limits, so building production traffic on them is a gamble. On OpenRouter the daily cap stays at 50 requests until you have topped up $10, and Hugging Face hands out free credits measured in dollars per month.
What it helps with
- pick an LLM provider with a permanent free tier
- check rate limits and region before wiring anything up
- repoint an OpenAI SDK client at another provider
- find an OpenAI-compatible endpoint for a prototype
- keep provider keys in environment variables
Example prompts for the agent
- Find me a free LLM API that allows at least 30 requests per minute
- Wire up Groq through the Anthropic SDK and show a minimal Python example
- Which of these free endpoints are Anthropic-compatible and which are not?
How to install awesome-free-llm-apis
1Skills CLI (Vercel)
npx skills add reason-machines/trending-skills --skill awesome-free-llm-apis -a claude-code2Manually: put the files in place
~/.claude/skills/awesome-free-llm-apis/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/awesome-free-llm-apis/
1Skills CLI (Vercel)
npx skills add reason-machines/trending-skills --skill awesome-free-llm-apis -a codex2Manually: put the files in place
.agents/skills/awesome-free-llm-apis/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add reason-machines/trending-skills --skill awesome-free-llm-apis -a cursor2Manually: put the files in place
.cursor/skills/awesome-free-llm-apis/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install reason-machines/trending-skills awesome-free-llm-apis2Manually: put the files in place
.github/skills/awesome-free-llm-apis/SKILL.md1Skills CLI (Vercel)
npx skills add reason-machines/trending-skills --skill awesome-free-llm-apis -a gemini-cli2Platform CLI
gemini skills install https://github.com/reason-machines/trending-skills --path skills/awesome-free-llm-apisManually: put the files in place
.devin/skills/awesome-free-llm-apis/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add reason-machines/trending-skills --skill awesome-free-llm-apis -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about awesome-free-llm-apis
How do I install awesome-free-llm-apis in Claude Code?
Run npx skills add reason-machines/trending-skills --skill awesome-free-llm-apis -a claude-code. The skill file lands at ~/.claude/skills/awesome-free-llm-apis/SKILL.md.
What does awesome-free-llm-apis do?
It gives the agent a reference on LLM providers with permanent free tiers: models, request limits, region, and OpenAI SDK integration samples in Python and TypeScript.
Are all listed providers OpenAI SDK-compatible?
No. Ollama Cloud is the exception. For the rest you only swap base_url and api_key in the client.
