FreeLLMMistral releases open weights for 1‑trillion‑parameter Large 4 model “Le Chonk”
Mistral makes its trillion‑parameter, multimodal Large 4 model available with open weights, targeting coding agents and cybersecurity.
Assista no YouTube · Assista ao Short
Mistral AI announced that the weights for its flagship model Mistral Large 4, nicknamed “Le Chonk,” are now publicly downloadable. The model has 1 trillion parameters overall and 52 billion active parameters, and it is natively multimodal with support for more than 160 languages.
The release follows three weeks of additional reinforcement learning that boosted performance on several benchmarks. The model scored 19.7% on Cognition’s FrontierCode 2.0, 48% on Artificial Analysis’s Coding Agent Index (matching GLM‑5.3 while using 47% fewer tokens), and 41% on Terminal‑Bench 4.0, a 13‑point rise from the preview.
In cybersecurity tests, Large 4 achieved 51% on the Artificial Analysis Cyber Index and refused 96.6% of malicious requests, the highest rate among open models.
Mistral positions Large 4 as a sovereign alternative to closed‑source models from OpenAI and Anthropic, especially for enterprises that need on‑prem AI for coding and security workloads.
O que isso significa para você
Open weights let organizations run a trillion‑parameter model inside their own firewalls, giving full control over data and inference costs. For users seeking free or locally hosted AI, Large 4 offers state‑of‑the‑art coding and cyber‑defense capabilities without relying on external providers, while still allowing commercial fine‑tuning under a permissive MIT‑style license.
No FreeLLMArena
Dos nossos rankings em 31 out 2026, baseados nas votações às cegas da Arena:
- Mistral Large (2411): #116 among free chat models
Fontes
- Mistral opens Large 4 weights for coding and security, TestingCatalog, 29 Oct 2026
- Mistral releases open weights for Mistral Large 4, providing sovereign control for the company’s ‘Le Chonk’ AI model, SiliconANGLE, 29 Oct 2026
- Mistral releases weights for its trillion-parameter AI model, targeting cybersecurity, TechCrunch, 29 Oct 2026
Escrito automaticamente a partir das fontes acima, depois verificado pelo nosso código: todo número e nome deve aparecer nelas, e nenhuma frase pode copiar uma fonte. Encontrou um erro? Reporte‑o.
Qual IA você pode usar grátis hoje?
Todo modelo de IA gratuito, classificado por votos humanos cegos, com a memória e velocidade que ele precisa na sua própria GPU.