Mistral Large 4 on FastMetal: 1T-parameter multimodal model in chat
Mistral Large 4 is available in FastMetal chat. Mistral AI released it as a public preview on October 6, 2026 and calls it its largest and most capable model to date.
FastMetal offers Mistral Large 4 in chat only. It is not available on the API or through MCP.
What it is
- A mixture-of-experts model with about 1 trillion total parameters and about 49 billion active per token, according to Mistral AI. Mistral says it was trained on text and images together from the start, as a native multimodal model.
- Built for coding and agent work. Mistral AI reports 61.7% on DeepSWE 1.1 and 28.3% on Terminal Bench 4.0. Those are Mistral's own figures; we have not reproduced them.
- Multilingual, with support for more than 160 languages by Mistral's count.
- A 524,288-token context and image input. Send a long document in one message, or attach an image and ask about it. We checked image reading with real images before launch.
- Open weights to follow. At launch, Mistral AI said it would publish the weights by the end of October 2026.
How it is billed
Mistral Large 4 is billed per million input tokens and per million output tokens. Our launch rate followed Mistral AI's preview pricing, and we revised it on October 20, 2026 when the preview pricing ended. Current rates are on the model page.
Billing is prepaid: you load a balance by card, from $5, and you can never spend more than you loaded. See pricing.
How to use it
- Create a FastMetal account.
- Load a balance.
- Open chat, pick Mistral Large 4 from the model list and ask your question. Attach images to a message as you need them.
Read its answers before you pick it
The model page shows Mistral Large 4's actual answers to our 43 shared prompts: summaries, bug fixes, translation and more. Put them next to the answers from the model you use today. Everything we serve is in the model catalog.