The Ultimate Guide to Free AI API Keys: 6 Platforms You Need to Know
If you build with AI, you need keys. This is the expanded version of the shortlist I actually use: one router for breadth, a few vendor consoles for latency and specialty models, GitHub when I already live in tokens, and Cloudflare when the edge is the product, plus how I combine them before I touch billing.
OpenRouter: one key reaches many models. Use the Models tab “free” filter so paid routes do not fail once you are out of credits; log the model ID on every request so 402s are obvious.
Google AI Studio, Groq, and GitHub expose usage in the console; NVIDIA and Cloudflare favor per-model keys or schemas. Plan for rate limits and label keys in your secret manager.
Groq Cloud lives at console.groq.com (not xAI Grok). Prefer expiring keys and revoke when a demo is done, with the same hygiene for GitHub PATs.
Treat “unlimited” UI rows as marketing-adjacent: upstream RPM/TPM and fair-use still apply; add backoff and a second provider before you demo to investors.
Production is not “free tier with fingers crossed”: paid plans buy predictable limits, support, and fewer surprises when a model is deprecated overnight.