AI

Introducing Mistral Large 4: Le chonk

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 06/10/2026, 20:27:25 UTC.

Briefings

Mistral released a preview of Mistral Large 4, a 1 trillion parameter model trained on 3,800 NVIDIA Grace Blackwell GPUs. The model supports two reasoning levels, with the 'high' level scoring 38 on Artificial Analysis, significantly outperforming Mistral Large 3's score of 9. It is positioned as a 6-month gap from the current frontier, though it remains far from Fable-class capabilities. The model is available via API with a specific 2,717 token output for the high level versus 3,275 for the none level.

Key points

• Mistral Large 4 is a 1 trillion parameter model trained on 3,800 NVIDIA Grace Blackwell GPUs, released via API with a 6-month gap from the current frontier.
• The 'high' reasoning level scored 38 on Artificial Analysis, significantly outperforming Mistral Large 3's score of 9, while the 'none' level scored 3275 tokens compared to 2717 for the high level.

Context and limitations

The evidence is limited to a single API preview available via Mistral's platform, with no public release date confirmed beyond the 'end of this month' promise, and no independent benchmarking data beyond the single Artificial Analysis score.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 07/10/2026, 04:05:25 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeibgzsqdgw75xesq5nriawclkud4eyid34nkg7c3iomuduvx6r3wqq