Skip to content
PAGAI Tools
#5 in LLM APIs & inferenceUsage-based

Groq

Fast inference for the open models Groq hosts.

Groq sells inference on its own hardware, with published per-token prices and high tokens-per-second figures for the models it hosts. The catalog is open models such as Llama, Qwen, and GPT-OSS, not every frontier model.

Official site

Why this rank

When the model you want is one Groq hosts and latency matters, Groq is the fast, simple API.

Best for

Low-latency responses from an open model.

Pricing

Usage-based, with published prices per million tokens. Groq says you can start free. Example rates on its pricing page include Llama 3.1 8B Instant at $0.05 per million input tokens and $0.08 per million output tokens. The catalog changes.

Pricing checked November 7, 2026. Confirm it on the official site.

Before you commit

  • You only get the models Groq serves.
  • A fast cheap model is not a substitute for a frontier model on a hard task.

Also look at

  • OpenRouter — One API key for many models and providers.
  • Gemini API — Google's Gemini models, with a free tier in AI Studio.
  • Mistral — European model lab with an API and open-weight models.