# Mistral Puts 1T-Parameter Large 4 on a Paid Preview API; Weights by Month-End, AA Scores 38

Times of AI Desk · 2026-10-06 · Models

[https://timesof.ai/2026/10/mistral-large-4-preview-1t-open-weights](https://timesof.ai/2026/10/mistral-large-4-preview-1t-open-weights)

> Mistral launched a paid Studio preview of Mistral Large 4 (ML4 / "le Chonk") on Oct. 6: a natively multimodal MoE with about 1.05T total / 49B active parameters, trained from scratch on 3,800 Nvidia Grace Blackwell GPUs in its European data centres. Mistral says it will release the weights by the end of the month; TechCrunch and The Next Web put that closer to Oct. 27. Artificial Analysis's first independent Intelligence Index score for the preview is 38 (#64 of 225), behind top open-weight models MiMo-V2.6-Pro (46), GLM-5.3 (45) and Kimi K3 (44). Most other benches are Mistral's own on a checkpoint it says will change.

Europe's open-weight lab just put a trillion-parameter-class model on a paid API — and gave cybersecurity partners a less-moderated build before the rest of the internet gets the weights. The first independent general score is not a crown.

**Mistral AI** on **October 6** opened a public preview of **Mistral Large 4** (**ML4**, nicknamed **"le Chonk"**) on **Mistral Studio**. The company says it **will release the weights by the end of the month**. TechCrunch and The Next Web report a date closer to **October 27**. Until then, Mistral says it is red-teaming with **cybersecurity leaders, vetted partners, and state authorities**, who get the same model **with reduced moderation and expanded cyber capabilities**.

## What shipped

Per Mistral's post and docs:

- A **natively multimodal** granular mixture-of-experts: Mistral's post rounds to **1T** parameters; docs list **1.05T total / 49B active**, plus a **1.6B** vision encoder. Listed as Public Preview, Open, v26.10.
- Trained **from scratch** on **3,800 Nvidia Grace Blackwell GPUs** in Mistral's own European data centres, which also serve the preview. VP Science **Pierre Stock** told TechCrunch the training used "only 4,000" GPUs — treat **3,800** as the primary figure.
- Target uses Stock named for TechCrunch: **cybersecurity, finance, and chip design**.
- Mistral calls ML4 the **first milestone funded by its €3B Series D** ([prior coverage](https://timesof.ai/2026/09/mistral-3b-euro-series-d)).

Context-window figures conflict: docs say **1M**; Artificial Analysis lists **524k** for the preview endpoint. Do not treat either as settled without saying which source.

## Claims vs checks

**Mistral-reported** benches on the preview checkpoint (Mistral says its RL run is still in flight):

| Bench | Score (Mistral) |
|------|-----------------|
| DeepSWE v1.1 | 61.7% |
| Terminal-Bench 4 | 28.3% |
| AutomationBench | 59.9% |
| Cybench | 93% |
| Lakera B3 (attack resistance) | 93.3% |
| Surge AI coding (blind, of 5) | 3.74 (2nd), behind Claude Opus 5 at 4.22 |

**Artificial Analysis** — first independent general read — lists **Mistral Large 4 Preview** at **38** on its Intelligence Index (**#64 of 225**), priced at **$1.36 / $4.18** per million input/output tokens. On the same board, top open-weight models are **MiMo-V2.6-Pro 46**, **GLM-5.3 (max) 45**, and **Kimi K3 (max) 44**. That puts the preview well behind the Chinese open frontier on this index — not a verdict on the final checkpoint.

Mistral says ML4 significantly outperforms any U.S. or European open-weight model and is competitive with the strongest open models globally. [Reflection's Beam](https://timesof.ai/2026/10/reflection-open-weight-imminent) is making a similar "best open outside China" pitch; **neither has public weights yet**. Re-test when the weights land.

## The release design

The sovereignty pitch is explicit: trained and served on European compute under European law. The contested design choice is the **pre-weights** access for governments and vetted security firms to a **less-moderated, more cyber-capable** build. Mistral argues closed models' refusals hurt defenders. That widens access to offensive-capable tooling; the cyber evaluations behind the pitch are largely Mistral's own. Attribute any "AA Cyber Index top five" claim to Mistral — it was not visible in the AA HTML we inspected.

## Limits

- Weights date: prefer Mistral's **"end of the month"**; attribute **Oct. 27** if using secondary outlets.
- Licence for the weights is **unconfirmed** in the post and docs we read. Do not call them Apache or "open source" until a licence ships.
- Nearly all benches except AA's Intelligence Index are **vendor-reported** on a **preview checkpoint**.
- Docs show a second, unlabelled price set at roughly half AA's figures ($0.68/$2.09); print **$1.36/$4.18** from AA unless Mistral labels the discount.
- Context: **1M (docs) vs 524k (AA)** — attribute before printing.

## Sources

- [Mistral AI: Introducing Mistral Large 4 (October 6, 2026)](https://mistral.ai/news/mistral-large-4/)
- [Mistral Docs: Mistral Large 4 (October 6, 2026)](https://docs.mistral.ai/models/mistral-large-4-0)
- [Artificial Analysis: Mistral Large 4 Preview (October 6, 2026)](https://artificialanalysis.ai/models/mistral-large-4)
- [Artificial Analysis: LLM Leaderboard](https://artificialanalysis.ai/leaderboards/models)
- [TechCrunch: Mistral's new 1T model aims to leapfrog closed and open rivals (October 6, 2026)](https://techcrunch.com/2026/10/06/mistrals-new-1t-model-aims-to-leapfrog-closed-and-open-rivals/)
- [Times of AI: Mistral's €3B Series D (September 8, 2026)](https://timesof.ai/2026/09/mistral-3b-euro-series-d)
- [Times of AI: Reflection Beam open-weight imminent (October 4, 2026)](https://timesof.ai/2026/10/reflection-open-weight-imminent)