Groq API pricing: per-token rates for fast inference
Checked Nov 6, 2026 on Groq’s official pages (sources below). Prices change often: confirm with the provider before you buy.
Groq runs its own LPU inference hardware and sells model access through GroqCloud, an OpenAI-compatible API billed per token (language models), per hour of audio (Whisper) or per character (text to speech).
groq.com/pricing redirected to the home page on Nov 6, 2026, so the figures below are quoted from the price column of GroqCloud’s supported models page (console.groq.com/docs/models). Some models are Enterprise only, with prices from sales.
Production models
- GPT OSS 120B (openai/gpt-oss-120b)
- Speed (tokens/sec)~500
- Price$0.15 input / $0.60 output per 1M tokens
- Context window131,072
- GPT OSS 20B (openai/gpt-oss-20b)
- Speed (tokens/sec)~1,000
- Price$0.075 input / $0.30 output per 1M tokens
- Context window131,072
- Llama 3.1 8B Instant
- Speed (tokens/sec)~560
- PriceEnterprise: contact sales
- Context window131,072
- Llama 3.3 70B Versatile
- Speed (tokens/sec)~280
- PriceEnterprise: contact sales
- Context window131,072
- Whisper Large V3
- Speed (tokens/sec)—
- Price$0.111 per hour of audio
- Context window—
- Whisper Large V3 Turbo
- Speed (tokens/sec)—
- Price$0.04 per hour of audio
- Context window—
Developer plan rate limits on the same page: GPT OSS 120B 250K tokens/min and 1K requests/min; Whisper Large V3 Turbo 400K audio seconds/hour.
Preview models (evaluation only)
- Qwen3.8 27B
- Price$0.80 input / $4.00 output per 1M tokens
- Safety GPT OSS 20B
- Price$0.075 input / $0.30 output per 1M tokens
- Canopy Labs Orpheus V1 English (TTS)
- Price$22.00 per 1M characters
- Canopy Labs Orpheus Arabic Saudi (TTS)
- Price$40.00 per 1M characters
- Llama Prompt Guard 2 22M / 86M
- Price$0.03 / $0.04 per 1M tokens (input and output)
Groq says preview models may be discontinued at short notice and should not be used in production.
Reading Groq’s prices
Language models are priced separately for input and output tokens, per 1M tokens. Whisper is priced per hour of audio and the Orpheus TTS models per 1M characters. The Llama models on the page show “Contact Sales” rather than a public rate. For other hosted open-model APIs see Together AI pricing and Replicate pricing; to put several providers behind one API see LLM routers.
Groq pricing FAQ
How much does the Groq API cost?
On Nov 6, 2026 GroqCloud listed GPT OSS 120B at $0.15 per 1M input tokens and $0.60 per 1M output tokens, and GPT OSS 20B at $0.075 / $0.30. Llama 3.1 8B and 3.3 70B were listed as Enterprise (contact sales).
What does Groq charge for Whisper transcription?
Whisper Large V3 is $0.111 per hour of audio and Whisper Large V3 Turbo $0.04 per hour, per console.groq.com/docs/models (Nov 6, 2026).
Is the Groq API OpenAI-compatible?
Yes. GroqCloud documents OpenAI compatibility and a models endpoint at api.groq.com/openai/v1/models.
Sources (official pages, read Nov 6, 2026)
Spot something out of date? Email hello@youraitools.co and we’ll fix it.