Mistral Releases Open Weights of 1T-Parameter Large 4, Tops FrontierCode Among Open Models
_AndrewZhao · x · 2026-10-29
Mistral open-sourced the weights of Le Chonk (Mistral Large 4), a 1T-parameter model targeting agentic coding and cybersecurity:
- Coding: #1 among open models on Cognition's FrontierCode 2.0 at 19.7% (DeepSeek V4.1 Flash 19.3%, GLM-5.3 17.8%, Kimi K3 17.3%, Beam 1.0 just 2.9%); leads Artificial Analysis Coding Agent Index (47) with strong token efficiency (8M tokens/task, in the "sweet spot"); wins 64% blind preference vs GLM-5.3 and Kimi K3 on Surge Code, but only 22% vs Claude Opus 5.5
- Security: 93.3% attack resistance on Lakera's B3 benchmark, validated by CrowdStrike, Cloudflare, Rubrik, Zscaler and others
- Deployment: weights on HuggingFace, customizable via Mistral Forge, self-hostable anywhere
Mistral touts major reasoning-efficiency gains that lower inference costs; third parties call it the best open-weight model from a Western lab.
More from Models
- Codex doubles usage limits for all Pro tiers after user backlash — pvncher · 2026-10-30
- Dream raises $17M seed from a16z to build spatial-reasoning AI for AEC — a16z · 2026-10-30
- Investor bets speech-to-speech will be massive, praises Phonic's model — graceisford · 2026-10-30
- User Praises OpenAI for Doubling Pro Usage and Including API in Pro Sub — Angaisb_ · 2026-10-30
- OpenAI doubles usage limits for Pro tiers and Business Premium, effective immediately — Polymarket · 2026-10-30
- Gemini 4 Argon tops 4 robotics tasks on RLE-Bench, ranks #2 behind GPT-6 Astra — daibond_alpha · 2026-10-30