Groq API pricing
7 models · USD per 1M tokens · verified 2026-10-23
Groq serves 7 of the models we track, 7 of them open-weights. The cheapest is Llama-3.1-8B-Instruct at $0.05 input and $0.08 output per million tokens. For 0 of these models, Groq is the cheapest tracked host.
// models on Groq
| model | lab | in $/1M | out $/1M | ctx | vs best price |
|---|---|---|---|---|---|
| Llama-3.1-8B-Instruct | Meta | $0.05 | $0.08 | 16.4K | +109.1% |
| GPT OSS 20B | OpenAI | $0.075 | $0.30 | 131.1K | +128.2% |
| GPT OSS Safeguard 20B | OpenAI | $0.075 | $0.30 | 128K | +28% |
| GPT OSS 120B | OpenAI | $0.15 | $0.60 | 131.1K | +273.4% |
| Llama 3.3 70B Turbo | Meta | $0.59 | $0.79 | 131.1K | +312.9% |
| Qwen3.6 27B | Alibaba (Qwen) | $0.60 | $3.00 | 262.1K | −11.1% |
| Qwen3.8 27B | Alibaba (Qwen) | $0.80 | $4.00 | 262.1K | +106.5% |
// how to read this page
Prices are Groq's base-tier list prices per million tokens. "vs best price" compares Groq's blended price (3 × input + output) / 4 with the lowest price we track for the same model, whether it comes from the lab's own API or another host. Open each model for its full price history and every host that serves it.