FreeLLMMistral releases Large 4 "Le Chonk" with 1 trillion parameters, strong coding performance
Mistral's new open‑source model runs on eight Nvidia B300 GPUs, scores high on coding benchmarks but lags Chinese rivals overall.
Paris‑based Mistral AI has published the weights for its largest model yet, Mistral Large 4, nicknamed "Le Chonk".
The model contains about one trillion parameters, of which 52 billion are active per inference step, and can process text and images to output text and code.
Inference requires eight Nvidia B300 GPUs for the standard version or four GB300 GPUs for a more compressed variant, according to the SGLang framework documentation.
Independent tests show Le Chonk leads open models on coding benchmarks, reaching 48 % on Artificial Analysis' Coding Agent Index while using 47 % fewer tokens than comparable models.
Mistral offers the model under a modified MIT license for free use, modification and commercial deployment, but commercial cloud providers must obtain a separate license.
Ce que cela signifie pour vous
Le Chonk gives developers a high‑performance, locally runnable coding model at a lower API cost, supporting the push for sovereign AI. However, its lower overall intelligence scores mean users seeking the best all‑round capability may still prefer Chinese open models or paid proprietary services.
Sur FreeLLMArena
D’après nos classements du 31 oct 2026, basés sur les votes à l’aveugle d’Arena :
- Mistral Large (2411): #116 among free chat models
Sources
- Paris-based Mistral releases "Le Chonk," a strong coding model that still trails China overall, The Decoder, 29 Oct 2026
Rédigé automatiquement à partir des sources ci‑dessus, puis vérifié par notre code : chaque nombre et chaque nom doivent y figurer, et aucune phrase ne doit copier une source. Une erreur ? Signalez‑la.
Quelle IA pouvez‑vous utiliser gratuitement aujourd’hui ?
Tous les modèles IA gratuits, classés par votes humains anonymes, avec la mémoire et la vitesse dont ils ont besoin sur votre propre GPU.