Mistral releases Large 4, a 1-trillion-parameter open-weight AI model with 49B active parameters
Mistral has released Mistral Large 4, codenamed "Le Chonk," a natively multimodal AI model with 1 trillion total parameters and 49 billion active parameters. The model is available via API and Mistral Studio, with open weights scheduled for release at the end of October. Mistral claims state-of-the-art performance on cyber defense, manufacturing, and finance workloads. Benchmark scores include 62% on DeepSWE v1.1, outperforming several competing models.
IllustrationEditorial responsibility
- No named human review is recorded for this page.
- Reports are grouped by semantic similarity and deterministic rules. Language models may assist titles, summaries, translation and cross-source analysis; the page reads the event directly, while its address stays stable when the title changes.
- Summary awaiting refresh
Summary awaiting refresh
Cross-source coverage
Reporting timeline
This is a surprising great release. Le Chaton fat is real! Mistral Large 4 puts Europe back in contention on coding and cybersecurity. Didnt expect Mistral to compete with GLM5.3! “Le Chonk” is a natively multimodal model
A post on X by user kimmonismus reports the release of Mistral Large 4, a natively multimodal AI model with 1 trillion total parameters and 49 billion active parameters, trained from scratch in European datacenters using only 4,000 NVIDIA Grace Blackwell Superchips. The post claims the model, nicknamed 'Le Chonk', achieves competitive benchmark scores: 61.7% on DeepSWE v1.1, roughly level with GLM-5.3's 61% but behind Kimi K3's 68%; a score of 50 on the Artificial Analysis Cyber Index, matching GLM-5.3-Flash and beating GLM-5.3's 36; and 82% on CyberGym-E2E-AA, ahead of MiMo-V2.6-Pro's 79%. The author expresses surprise that Mistral can compete with GLM-5.3 and concludes that competing with much less compute is possible, congratulating Mistral.
Read sourceMistral Releases Mistral Large 4: 1T Parameters, 49B Active, Native Multimodal Model
Mistral has announced the release of Mistral Large 4, a native multimodal model with 1 trillion total parameters and 49 billion active parameters. The company claims it is the best open-weight model from the US or Europe. According to the announcement, the model achieves state-of-the-art performance on critical workloads including cyber defense, manufacturing, and finance. Its visual grounding capabilities reportedly surpass closed-source frontier models. The model has been trained and is served on Mistral's own computing infrastructure, with an API now open. Mistral is privately collaborating with cybersecurity partners. The open-weight version is scheduled for release at the end of October. CEO Arthur Mensch added that reinforcement learning shows no signs of saturation.
Mistral releases Mistral Large 4 open model with 62% DeepSWE score, 1T parameters
In a post on X, user DanDr1s announced that Mistral has shipped Mistral Large 4, claiming it is the best open model outside China. The post includes benchmark results on the DeepSWE evaluation, showing Mistral Large 4 scoring 62%, ahead of GLM 5.3 at 61%, DeepSeek V4 Pro at 57%, and Qwen 3.8 Max at 51%. The model is described as having 1 trillion total parameters with 49 billion active parameters. The post states that model weights will be released on October 27. This announcement positions Mistral Large 4 as a leading open-weight AI model in the competitive landscape outside of China.
Read sourceShow 2 older updatesHide older updates
Mistral Releases Large 4 Public Preview, 1 Trillion Parameter Model Weights to Open
Mistral has released the public preview of Mistral Large 4, codenamed 'le Chonk', a natively multimodal AI model with 1 trillion total parameters and 49 billion active parameters. The model is available for trial on Mistral Studio. According to the announcement, the model's weights will be made open by the end of this month. The company has published benchmark scores across multiple domains, allowing observers to assess the model's position within the open-source AI landscape. The release marks a significant step in large-scale open-weight AI models.
Read sourceMistral announces Mistral Large 4 'Le Chonk', a 1T-parameter open-weight AI model
Mistral has announced Mistral Large 4, codenamed 'Le Chonk', a new 1-trillion-parameter open-weight AI model. The model features 49 billion active parameters and native multimodality. It is rolling out via APIs today, with an open-weight release planned for the end of October. Mistral claims the model achieves state-of-the-art performance on critical workloads, including cyber defense. The announcement was made via a post on X, with a link to further details.
Read source