Groq

Groq runs open models on its LPU hardware for very low-latency inference behind an OpenAI-compatible API. Supports chat, tool use and fast Whisper speech-to-text; works with the groq SDK or the OpenAI SDK via a base URL.

Website Docs Changelog Compare with others

Groq at a glance

From official pages, checked Sep 29, 2026. Tap the arrow for each source.

Free tier: Yes
Free plan with per-model rate limits.
Starting price: $0.15 per 1M input tokens
GPT-OSS 120B; $0.60 per 1M output tokens.
Rate limit: 30 requests/min
Free plan, openai/gpt-oss-120b; 1K requests/day, 8K tokens/min.
Size limit: 131K token context
GPT-OSS 120B (Whisper speech-to-text: 100 MB max file).
Official SDKs: PythonJavaScript / TypeScript
EU hosting: No
Retained customer data is stored in GCP buckets in the United States.
Compliance: SOC 2HIPAA

What you can build with Groq

GroqSnippetPaste

Transcribe audio files with Whisper Large v3 Turbo on Groq

Send an audio file to Groq's whisper-large-v3-turbo with the groq Python SDK and get the transcript back.

Free tier: YesFrom $0.15 per 1M input tokens

Voice

Groq at a glance

What you can build with Groq

Transcribe audio files with Whisper Large v3 Turbo on Groq

Update timeline