[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f3li4p7o7rjii8":3},{"code":4,"message":5,"data":6,"timestamp":39},10000,"success",{"full_name":7,"position":8,"description":9,"language":10,"stars":11,"forks":12,"stars_gained":13,"age_days":14,"hn_threads":15,"one_liner":16,"summary_md":17,"topic":18,"github_topics":21,"license":35,"homepage":36,"updated_at":37,"history":38},"mnfst\u002Fawesome-free-llm-apis",11,"List of Permanent Free LLM API (API Keys)","JavaScript",11171,1251,2968,217,[],"A curated list of LLM API providers with permanent free tiers, for developers who need a zero-cost key now.","## The problem\n\nYou want to prototype against a real LLM without a credit card, or you want a fallback provider when your paid quota runs dry. Finding out which providers actually offer a standing free tier — and what the limits are — means trawling pricing pages, signup flows and forum threads that go stale within weeks. This repository answers that with a single maintained table: provider, base URL, models, context window, modality and rate limits, with each link pointing at the key-issuing page.\n\n## What's different\n\nMost \"free API\" resources are blog posts that rot, or they mix trial credits, expiring promos and genuinely permanent tiers without saying which is which. This one separates providers that train their own models from third-party inference hosts, and states the constraint that matters most in practice — no credit card, non-commercial only, prompts used for training, phone verification — inline with each entry. It also flags OpenAI SDK compatibility, so an endpoint drop-in works with existing client code. That editorial discipline is the product; there is no code, no SDK and no router here despite the repo's topics.\n\n## Who it's for\n\nDevelopers, students and hobbyists building prototypes, agents or evaluation harnesses on a budget, and anyone who wants a second provider wired in for redundancy. It is useful for infrastructure-minded people who need to know rate ceilings before designing around a provider. Look elsewhere if you need guaranteed throughput, SLAs or commercial-use terms — several listed tiers explicitly prohibit commercial use or reserve prompts for training, so this is not a procurement list. Teams shipping production traffic should treat it as a discovery index, not a dependency.\n\n## Cost of adoption\n\nThere is nothing to install: it is a README, CC0-1.0 licensed, so you can copy entries freely. The real cost is operational — free tiers change without notice, as one entry in the list itself shows with a retirement already announced, and limits differ by orders of magnitude between providers. You will spend time on signup flows, phone verification and per-provider SDK quirks. Verify each provider's current terms yourself before relying on it; a listing here is not a guarantee. For the sponsored link and the provider blurbs, apply normal skepticism.\n\n## Verdict\n\nThe project is roughly seven months old and pulled about 3,000 stars in a single month on the monthly JavaScript board, with no Hacker News discussion and no daily trending appearances — a fast, recent surge rather than a long track record. Its value scales with maintenance: a stale list is worse than no list, and the README is honest enough to mark announced retirements, which is a good sign. Use it as a starting point for cheap experimentation, cite it when you write your own fallback chain, and re-check the numbers before you commit. Not a library, and not trying to be.",{"slug":19,"name":20},"data","Data",[22,23,24,25,26,27,28,29,30,31,32,33,34],"ai-agents","anthropic","awesome","awesome-list","gemini","llm","llm-router","llm-routing","ollama","openai","openclaw","openclaw-plugin","router","CC0-1.0",null,"2026-10-24T16:36:30Z",[],1792868684]