← Founder Notes
Archive

The sovereign model just went big. mistral's large 4, nicknamed chonk, is a 1.05-trillion-parameter…

Yethikrishna ROriginal on Threads

the sovereign model just went big. mistral's large 4, nicknamed chonk, is a 1.05-trillion-parameter moe with 49 billion active, trained on 4,000 nvidia grace blackwell gpus in european data centers, with open weights out oct 27.

the frontier now has a passport.

Context

Mistral launched a public preview of Mistral Large 4 on Oct 6, 2026, nicknamed le Chonk, a Mixture-of-Experts model with 1.05 trillion total parameters that is natively multimodal. Mistral says weights drop at the end of the month.

DataCamp reports it was trained from scratch on Nvidia Grace Blackwell GPUs in Mistral's own European data centers.

How it compares

The 1.05 trillion parameters, the chonk nickname, the Grace Blackwell training and the European data centers match the sources. The active parameter count differs: Mistral's own post and docs say 52 billion, while DataCamp says 49 billion. DataCamp gives 3,800 GPUs, not 4,000. The Oct 27 weights date was not seen in the excerpts read, so unsupported here, not refuted. Mistral says only end of this month.

'the frontier now has a passport' is the author's opinion.

Related work

Watch next

  • Check Mistral's model card for the active parameter count.

Sources

  1. Mistral, Oct 6, 2026mistral.ai
  2. Mistral Docsdocs.mistral.ai
  3. DataCampdatacamp.com

Provenance

The note above is reproduced unedited from the original post, first published on Threads on 10 October 2026 at 21:40 IST. Sources are the papers and datasets the note draws on.

View the original post
Embed this note
<iframe src="https://founder.myndlabs.tech/notes/embed/the-sovereign-model-just-went-big-mistral-s-DeUewtACBuN" width="480" height="420" style="border:0;max-width:100%" loading="lazy" title="The sovereign model just went big. mistral's large 4, nicknamed chonk, is a 1.05-trillion-parameter…"></iframe>

More notes