Mistral Large 4 (AI model): news timeline & CVEs

Mistral says "Le Chonk" can challenge the best AI models

Mistral released Le Chonk, a 1-trillion-parameter open-weight model it says rivals leading US and Chinese systems.

French lab Mistral previewed Mistral Large 4, nicknamed Le Chonk, a freely available 1-trillion-parameter model it says was trained from scratch and approaches top proprietary systems. The model is optimized for coding, cyberdefense, manufacturing, finance, and electrical engineering, with a final release expected by month's end. Mistral also cited a September $3.3 billion round at a $24 billion valuation and said open-weight use costs only the compute consumed.

Ars Technica · AIOct 7, 14:12 UTC · 1d agoModel release1· 1 read

[AINews] Quasi-Riemann-Hypothesis: OpenAI publishes 722 math papers solving 90 of the top 500 open math problems; “the most significant moment” in >100 years of mathematics

OpenAI published 722 math manuscripts from an unreleased model, claiming progress on many open problems including a quasi-Riemann result.

OpenAI published 722 mathematical manuscripts in 372 families from an unreleased internal model, based on about 4,000 problems and roughly three hours of ChatGPT Pro compute per result, after consulting the Institute for Advanced Study. Commentators highlight a quasi-Riemann hypothesis result, faster-than-n-log-n integer multiplication, and a 3D elastic inverse uniqueness claim open since 1994, but the results are unverified and some researchers expect errors. The same roundup covers Mistral Large 4, a 1-trillion-parameter multimodal preview with 49 billion active parameters, priced at $1.36/$4.18 per million tokens and trained on about 3,800 Grace Blackwell GPUs in Europe, with open weights promised for the end of October.

Latent SpaceOct 7, 04:55 UTC · 1d agoAI research1

llm-mistral 0.16

llm-mistral 0.16 adds LLM CLI support for Mistral reasoning models, including Mistral Large 4.

Simon Willison released llm-mistral 0.16, a plugin for the LLM command-line tool. The update adds support for Mistral reasoning models, including the newly released Mistral Large 4. The note is tagged llm, mistral, and llm-reasoning.

Simon WillisonOct 6, 21:32 UTC · 2d agoAI tools & infra1

Introducing Mistral Large 4: Le chonk

Mistral previewed Mistral Large 4, a 1-trillion-parameter model with 49 billion active parameters.

Mistral released a preview of Mistral Large 4, a 1 trillion parameter model with 49 billion active parameters. It was trained on Mistral's cluster of 3,800 NVIDIA Grace Blackwell GPUs and is available through the Mistral API, with open weights promised by the end of the month. The API supports only two reasoning levels, "none" and "high". On Artificial Analysis it scores 38, just behind DeepSeek 4.1 Flash, a 552B model.

Simon WillisonOct 6, 20:18 UTC · 2d agoModel release 8 sources

Mistral Large 4

Simon Willison compares Mistral Large 4 with Claude, GPT, and Gemini on a whimsical SVG prompt.

Simon Willison commented on a Hacker News thread about Mistral Large 4, responding to a claim that standard benchmarks are saturated. Using his llm CLI at default reasoning levels, he asked Claude Opus 5.5, GPT-6.1 Sol, Gemini 3.8 Flash, and mistral/mistral-large-4 to generate the same SVG of an armadillo in fishnet tights jaywalking on Mars. The post is a light model comparison and does not include release specs, parameter counts, or benchmark scores.

Simon Willisonupdated Oct 6, 20:18 UTC · 2d agofirst Oct 6, 18:20 UTC · 2d agoModel release 8 sources1

Mistral AI Releases Mistral Large 4 (Le Chonk): A 1.05T Parameter Multimodal MoE Model

Mistral released Mistral Large 4, a 1.05T multimodal MoE, with open weights due in late October.

Mistral AI released Mistral Large 4, nicknamed Le Chonk, as a public preview. It is a 1.05 trillion parameter Mixture-of-Experts model with 49 billion active parameters, native image input, and a 1 million token context window. Training used 3,800 NVIDIA Grace Blackwell GPUs in Mistral's European datacenters. The API is available now, and open weights are scheduled to ship at the end of October 2026.

MarkTechPostupdated Oct 6, 20:18 UTC · 2d agofirst Oct 6, 17:30 UTC · 2d agoModel release 8 sources1

Mistral’s new 1T model aims to leapfrog closed and open rivals

Mistral AI released Mistral Large 4, a one-trillion-parameter multimodal model, with open weights promised after safety testing and cyber/finance optimization.

Mistral AI launched Mistral Large 4 (ML4), nicknamed 'Le Chonk,' a one-trillion-parameter multimodal model currently available only via a guarded API endpoint, with weights planned for release in about three weeks after safety testing. The model was trained entirely on Mistral's own compute using 4,000 NVIDIA GPUs, which the company says is two to three times less than Chinese competitors. Optimized use cases include cybersecurity, finance, and chip design; backers ASML and Samsung led recent rounds valuing the company at €21 billion (about $24.39 billion). Independent benchmark results are still pending.

TechCrunch · AIupdated Oct 6, 20:18 UTC · 2d agofirst Oct 6, 14:33 UTC · 2d agoModel release 8 sources

Mistral Large 4 is Europe's trillion-parameter answer to US models that refuse security work

Mistral Large 4 scores 38 on Artificial Analysis Intelligence Index, leading non-Chinese open models but trailing Claude Opus 5.5 by 20 points.

Mistral Large 4 (1 trillion parameters, 49 billion active) scores 38 on the Artificial Analysis Intelligence Index, up sharply from Mistral Large 3's 9, edging past GLM-5.2 but trailing leader Claude Opus 5.5 at 58. Mistral claims top-five placement on the Cyber Index and an 82% score on a vulnerability reproduction-and-patch test, noting Claude Opus 5.5 and GPT-6 Astra score near zero because they refuse the task. Weights are expected end of October after red-teaming with security firms, partners, and governments; training ran on 3,800 Nvidia Grace Blackwell GPUs in European data centers funded by a €3 billion Series D. ML4 also takes second place in coding (49.8% on the Coding Agent Index) ahead of Deepseek V4 Pro and Qwen3.8 Max.

The Decoderupdated Oct 6, 20:18 UTC · 2d agofirst Oct 6, 14:18 UTC · 2d agoModel release 8 sources

Mistral Large 4: "Le Chonk"

Mistral previews Mistral Large 4, a 1-trillion-parameter open-weight multimodal model, with weights due by month's end.

Mistral launched a public preview of Mistral Large 4, nicknamed Le Chonk, a natively multimodal model with 1 trillion parameters and 49 billion active. Open weights are scheduled by month's end, and the preview API is available on Mistral Studio. It was trained from scratch on 3,800 NVIDIA Grace Blackwell GPUs in Mistral's European datacenters. Reported results include 82% on an Artificial Analysis vulnerability reproduction-and-patch test, 93% on Cybench, and a 49.8% Coding Agent Index.

Hacker News · AIupdated Oct 6, 20:18 UTC · 2d agofirst Oct 6, 13:25 UTC · 2d agoModel release 8 sources HN 65↑ · 6 comments

Mistral Large 4

Mistral released Mistral Large 4, an open-weight multimodal MoE with 1.05T total parameters.

Mistral published Mistral Large 4 in public preview (v26.10) on October 6, 2026. It is an open-weight, general-purpose multimodal Mixture-of-Experts model with 49 billion active parameters, 1.05 trillion total parameters, and a 1.6 billion-parameter vision encoder. The release lists a 1 million-token context, API pricing, and support for structured outputs, function calling, document Q&A, agents, and built-in tools.

Hacker News · AIupdated Oct 6, 20:18 UTC · 2d agofirst Oct 6, 13:15 UTC · 2d agoModel release 8 sources HN 175↑ · 40 comments

Mistral Large 4

A Hacker News thread on Mistral Large 4 has 46 points and four comments.

A Hacker News item titled Mistral Large 4 shows 46 points and four comments. The supplied text does not include model size, benchmark scores, availability, or weight-release details.

Hacker News · AIupdated Oct 6, 20:18 UTC · 2d agofirst Oct 6, 13:06 UTC · 2d agoModel release 8 sources HN 46↑ · 4 comments

Mistral Large 4

Timeline11 articles

Appears with

Mistral says "Le Chonk" can challenge the best AI models

[AINews] Quasi-Riemann-Hypothesis: OpenAI publishes 722 math papers solving 90 of the top 500 open math problems; “the most significant moment” in >100 years of mathematics

llm-mistral 0.16

Introducing Mistral Large 4: Le chonk

Mistral Large 4

Mistral AI Releases Mistral Large 4 (Le Chonk): A 1.05T Parameter Multimodal MoE Model

Mistral’s new 1T model aims to leapfrog closed and open rivals

Mistral Large 4 is Europe's trillion-parameter answer to US models that refuse security work

Mistral Large 4: "Le Chonk"

Mistral Large 4

Mistral Large 4