Mistral releases Large 4 open weights, nicknamed Le Chonk
Original title: Paris-based Mistral releases "Le Chonk," a strong coding model that still trails China overall
The trillion-parameter model leads some open coding scores but trails top Chinese models and Claude overall.
Mistral published Mistral Large 4 weights on Hugging Face three weeks after a public preview. The company says further reinforcement learning raised Terminal-Bench 4.0 from 28 to 41 percent and put it near GLM-5.3 on a coding-agent index while using fewer tokens. Artificial Analysis scores the final high mode 26 on Intelligence Index v5, below DeepSeek V4.1 Flash and GLM-5.3 at 32. API pricing is $0.70 per million input tokens and $2.10 per million output tokens under a modified MIT license.
Unconfirmed: Mistral calls Large 4 the leading open cyber-defense model, but its chart shows DeepSeek and MiMo ahead.
- About one trillion parameters, with 52 billion active per step.
- SGLang docs say the standard build needs eight Nvidia B300 GPUs.
- Mistral says it refuses 96.6 percent of malicious cyber requests.
- Companies offering Large 4 as a commercial cloud service need another license.
- Unreleased Kimi K3.1 Pro scores 67 percent on Terminal-Bench 4.0.
- Vendors
- Mistral
- Organizations
- Artificial AnalysisHugging FaceOrange Cyberdefense
- Countries
- France
From the original article
6 min read at the-decoder.com
Mistral has released the weights for Large 4, aka "Le Chonk," a trillion-parameter model trained entirely in Europe. Mistral says it beats its open rivals at coding, and independent testing shows it's fast and fairly cheap. It still can't match China's top models, scoring 26 points to the 32 posted by DeepSeek V4.
By Maximilian Schreiner · the-decoder.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.