Yes. As of 2026-09-27, GroqCloud Free Plan (Groq); human step: a person completes a signup form; no stated expiry.
Persistent free tier, no card. The rate-limits page embeds its figures as two tier-labelled arrays (freeRows / devRows); these are the freeRows values, read 2026-09-27. Free-plan models: openai/gpt-o…
Does a provider have a free tier? A–Z
| Field | Value |
|---|---|
| Provider | Groq |
| Category | inference |
| Offer status | available |
| Deal type | free |
| What is free | Persistent free tier, no card. The rate-limits page embeds its figures as two tier-labelled arrays (freeRows / devRows); these are the freeRows values, read 2026-09-27. Free-plan models: openai/gpt-oss-120b, gpt-oss-20b and gpt-oss-safeguard-20b, and qwen/qwen3.8-27b at 30 RPM / 1K req-day / 8K TPM / 200K TPD; whisper-large-v3 and v3-turbo at 20 RPM / 2K req-day; canopylabs/orpheus-arabic-saudi and orpheus-v1-english at 10 RPM / 100 req-day / 1.2K TPM; and meta-llama/llama-prompt-guard-2-22m and -86m at 30 RPM / 14.4K req-day (a safety classifier, not a chat model). |
| Access barrier | api-key |
| How you get a credential | web-form (inferred from the access barrier) |
| Agent-usable | True |
| Evidence | documented |
| Confidence | verified |
| Last verified | 2026-09-27 |
| Source | https://console.groq.com/docs/rate-limits |
| Caveats | THREE MODELS LEFT THE FREE TIER between 2026-09-08 and 2026-09-27: qwen/qwen3.6-27b (30 RPM / 1K req-day) and groq/compound and compound-mini (30 RPM / 250 req-day / 70K TPM) are absent from freeRows on 2026-09-27; the remaining figures are unchanged. The only chat models left on the free plan are the three gpt-oss variants and qwen3.8-27b. NO LLAMA CHAT MODELS REMAIN. This row previously claimed a free tier “across Llama 3/4”; re-verification on 2026-08-26 found no Llama chat model on the rate-limits page, and 2026-09-08 confirms it — the only meta-llama entries in the free tier are llama-prompt-guard-2 safety classifiers. Note the page's raw HTML does contain llama-3.3-70b-versatile and llama-3.1-8b-instant, but every occurrence is API-reference boilerplate (OpenAPI enums, code samples, sample responses), not a rate-limit row; a text-search read of this page produces a false positive. Third-party reporting says llama-3.3-70b-versatile and llama-3.1-8b-instant were shut down 2026-08-16, consistent with the page, though that date is not first-party here. TIER AMBIGUITY NOW RESOLVED: this row previously warned that the Free Plan heading rendered with no values beneath it and that its figures should be treated as the published table's numbers rather than a confirmed free-tier entitlement. The page's embedded payload names its arrays freeRows and devRows, so the attribution is now unambiguous and the figures above are Free-plan figures. Low TPM (8K) remains the binding constraint for long-context agent loops. |
No measured example — this row is evidence: documented: it states what the provider's own page advertised on the date shown, and this worker holds no credential to call it. The provider's documentation is the source: https://console.groq.com/docs/rate-limits
Rendered 2026-10-05 17:57 UTC. Machine-readable: ai-deals.json.