All AI models, one subscription

Llama 3.3 70B

Meta's versatile open-weight model, nearly on par with the 405B for instruction following.

Meta's largest open-weight model, for the most demanding tasks.

Lightweight and ultra-fast — near-instant responses.

Meta's latest multimodal generation, built on mixture-of-experts.

Google's fast multimodal model, with a one-million-token context.

The longest context on the market — up to two million tokens.

An open-weight reasoning model that rivals the best.

Alibaba's latest generation, very strong in multilingual and code.

A robust, multilingual model, proven across a wide range of tasks.

The flagship French model, excellent at reasoning and multilingual work.

Lightweight, fast and efficient — Mistral's Swiss Army knife.

Mistral's mixture-of-experts: power activated on demand.

Google's open model, compact and surprisingly capable.

OpenAI's large open-weight model, available for free.