Mistral's biggest open model is on its API now, with downloadable weights due this month
On 6th Oct'26, Mistral put Large 4 into public preview on its API at $1.36 per million input tokens and $4.18 per million output tokens. It is a mixture-of-experts model with 1 trillion parameters, of which 49 billion work on any given request, it reads images as well as text, and Mistral trained it on about 3,800 Nvidia Grace Blackwell GPUs in its own European data centres. The downloadable weights are due at the end of October, after red-teaming with security partners and state authorities, and Mistral says the model is still in reinforcement-learning training during the preview. Its own benchmarks put it ahead on some legal, finance and visual tests, and it acknowledges the model still trails the top closed models in areas such as coding.
What it actually means
The open-weight choice for a serious model has mostly meant a Chinese lab, and for a bank in Riyadh or a ministry in Brussels that origin is a procurement question before it is a technical one. Large 4 gives those buyers a model trained and served in Europe by a European company, with weights they can host on their own servers once released, so a regulated client who said no to DeepSeek or Qwen on provenance has something new to say yes to.
Hold off on building anything on the preview, though, because Mistral says the model is still being trained and the version you test this week may not be the one that ships. Run your evaluation set against it now to see where it lands, then run it again when the weights arrive at the end of October and decide on the second result.
A regulated client who said no to Chinese open weights on provenance now has a European option to test.
Source: Mistral · Wednesday 7 October 2026