Groq quietly updated its free-tier limits this week. No blog post, just a changed number in the docs — which is exactly the kind of thing this site exists to catch.
What changed
Requests per minute on the popular open-weight models dropped by roughly half. Daily token caps stayed the same. Existing API keys keep working; nothing was revoked.
What it means for you
If you were prototyping: you will barely notice. If you were running a small production workload on the free tier: this is your warning shot. Free tiers that get popular get trimmed — budget for the paid tier or add a fallback provider now, not after the next cut.
Still the fastest free option
Even halved, Groq's inference speed remains the best free tier for latency-sensitive demos. Our guide to free LLM APIs has the updated numbers and alternatives.
