Mistral News 10/29/2026 AI Rating: High

Mistral releases open weights for Mistral Large 4 (Le Chonk)

#Mistral#Open weights#Coding#Mistral Large 4

Mistral released the open weights of Mistral Large 4, nicknamed Le Chonk, on October 29, 2026, following a public preview earlier in the month. It is a 1-trillion-parameter, natively multimodal model trained from scratch on European infrastructure, which Mistral calls the best open-weight model for coding.

Details

  • Coding: 19.7% on Cognition’s private FrontierCode 2.0; 48% on the Artificial Analysis Coding Agent Index (roughly tied with GLM-5.3, with 47% fewer tokens per task); 41% on Terminal-Bench 4.0, up 13 points from the preview. In Surge Code blind evaluations it ranked behind only Claude Opus 5.5.
  • Reasoning modes: Low, Medium and High (the default).
  • Agents and work: 73.3% on FinWorkBench, second to Kimi K3 among open models; 84.5% on BrowseComp; resists 97.1% of attacks on the B3 Agent Security Benchmark.
  • Cybersecurity: 51% on the Artificial Analysis Cyber Index and an average 96.6% refusal rate for malicious cyber requests. Mistral says it is modifying guardrails after seeing potential uplift for real attacks, and a penetration-testing variant is in vetted preview with partners such as Orange Cyberdefense.
  • License: Mistral Ecosystem Open-Weight license, a modified MIT license. Model-as-a-Service operators need a separate commercial license.
  • Availability: Mistral Studio, the Mistral API, OpenRouter, Hugging Face (Mistral-Large-4-1T-A52B, for vLLM or SGLang), NVIDIA endpoints and NIM, Microsoft Foundry and Vibe Code.
  • Pricing: $0.70 per million input tokens, $0.07 cached input, $2.10 output.
  • Technical report: Described as coming soon.

What happened next

The weights release turns the October preview into something teams can self-host and fine-tune, with three more weeks of RL behind the gains. Mistral says Large 4 will be the base for more specialized models. The scores are vendor-reported; the technical report has not yet been published.

Source https://mistral.ai/news/mistral-large-4-oss/

Related Articles

More Mistral articles →
Mistral News 10/06/2026 Read AI Rating: High

Mistral Launches Mistral Large 4, a 1-Trillion-Parameter Open-Weight Model Nicknamed 'Le Chonk'

Mistral AI unveiled Mistral Large 4, a natively multimodal 1-trillion-parameter Mixture-of-Experts model trained on 3,800 Nvidia Grace Blackwell GPUs, with API preview access now and full open weights due October 27.
Mistral News 11/05/2026 Read AI Rating: Medium

Mistral expands Studio with managed agents, workflows, and apps for Le Chonk

Mistral Studio adds managed agents with skills, sandboxes, subagents, and hooks, plus workflows with event triggers, enterprise-ready apps, a Mistral MCP server, OpenTelemetry traces, and evaluations.
Mistral News 10/20/2026 Read AI Rating: High

Mistral Launches Voxtral 3 Speech Models and a Real-Time Voice Agent SDK

Mistral AI released Voxtral 3, a family of speech-to-text and text-to-speech models plus a voice agent SDK that runs listening, reasoning, tool calling and speaking as one streaming pipeline.
Mistral News 10/15/2026 Read AI Rating: Medium

Mistral Launches Timestral, Its First Time-Series Foundation Model Family

Mistral AI introduced Timestral, a family of four probabilistic, multivariate time-series forecasting models (5M-209M parameters) that top the GIFT-Eval, TIME, and fev-bench leaderboards and run on CPU without a GPU.