Mistral 开放 Mistral Large 4 权重
Mistral AI 开放了 Mistral Large 4(又称 Le Chonk)的模型权重。这是一款原生多模态、多语言 MoE 模型,拥有约 1 万亿总参数、490 亿激活参数和 100 万 Token 上下文窗口。Hugging Face 提供 FP8 与 NVFP4 官方版本,Unsloth 和 llama.cpp 团队还制作了社区 GGUFs。模型采用 Mistral Ecosystem Open-Weight License,允许使用、修改、微调和商业使用,但模型即服务业务需另行取得商业许可。Mistral 表示该模型领先开放权重模型的编程与防御性安全基准,并提供三种推理模式。报道列出的外部评测结果包括:Cognition 私有基准 FrontierCode 2.0 得分 19.7%;Artificial Analysis Coding Agent Index 得分 48%,与 GLM-5.3 持平,平均每项任务使用的 Token 少 47%。
此次发布让开发者可以使用一款万亿参数级别的开放权重多模态模型,并获得官方低精度格式、本地运行和自托管选项,以及每百万输入 Token 0.70 美元、缓存输入 0.07 美元、输出 2.10 美元的 API 定价。
Thinking
Summary
We release the weights of Le Chonk, advancing the frontier of open models in agentic coding and cybersecurity.
Earlier this month, we released Mistral Large 4 (Le Chonk), our flagship 1-trillion-parameter, natively multimodal model, in public preview. The response to the preview has exceeded our expectations, and we're grateful to our developer community, early-access customers, and partners for their support, collaboration, and feedback.

Since then, we've put that feedback to work while continuing to scale reinforcement learning, driving massive improvements in coding and defensive cybersecurity. Gains continue to accelerate, with no signs of saturation.
#Mistral AI#Mistral Large 4#Le Chonk#Mistral Ecosystem Open-Weight License#FP8#NVFP4#Unsloth#llama.cpp