Thinking ยท mentor.work ยท AI view
๐ Turn ~800M Free AI Tokens Into a Single OpenAI API
Just discovered a really clever open-source project: FreeLLMAPI
The idea is simple but the execution is clean — aggregate free tiers from 14 major AI providers:
Google Gemini · Groq · Cerebras · SambaNova · NVIDIA · Mistral · OpenRouter · GitHub Models · Hugging Face · Cloudflare · Zhipu · Moonshot · MiniMax…
→ into one single endpoint, fully compatible with the OpenAI SDK.
What I love most:
Not the ~800M tokens/month number. It's how it solves a very real problem — you have 14 free tiers, but managing 14 different SDKs is something nobody actually does.
FreeLLMAPI does exactly one thing well: collapse everything into one familiar interface.
Point your OpenAI SDK at localhost:3001/v1 and you're done. Zero code changes.
Technical highlights worth noting:
- Automatic failover — 429 / timeout / error → retries the next provider, up to 20 attempts
- Sticky sessions — multi-turn conversations stay on the same model for 30 minutes, avoiding hallucination spikes from mid-conversation model switches
- Per-key rate tracking — monitors RPM/RPD/TPM/TPD per key to stay under every free-tier cap
- AES-256-GCM encryption — keys encrypted before hitting SQLite
- Runs on a Raspberry Pi 4, ~40MB RAM at idle
But let's be honest:
Endpoint intelligence degrades throughout the day — Gemini 2.5 Pro and GPT-4o have the lowest daily caps. Once exhausted → fallback to weaker models. Resets at UTC midnight.
Tool calling and vision are not yet supported. Strictly personal use — don't expose it to the internet.
This is a tool for prototyping and learning, not production.
3-line setup:
git clone https://github.com/tashfeenahmed/freellmapi
cd freellmapi && npm install && npm run dev
Open localhost:5173, add your provider API keys, grab your unified key → done.
Perfect for anyone building AI agents, coding assistants, or just experimenting without spending money upfront.
GitHub → github.com/tashfeenahmed/freellmapi
#AI #LLM #OpenSource #SelfHost #FreeTier
This article was AI-assisted and edited by Mervin. All facts were verified against primary sources before publishing.