Mistral Large 4 预览上线：1T 总参数 / 49B 激活、自建 3800 张 Grace Blackwell 集群训练，Artificial Analysis 38 分重回距前沿约半年之列，月底开放权重
Introducing Mistral Large 4: Le chonk
精选评分 76
导读
Mistral 发布 Mistral Large 4 预览：一万亿总参数、49B 激活参数，在其自有的 3,800 张 NVIDIA Grace Blackwell GPU 集群上训练；API 即日可用，并承诺“本月底”放出开放权重。模型只暴露 none 与 high 两档推理，high 效果更好而输出 token 反而更少（2,717 对 3,275）。Artificial Analysis 得分 38、略低于 552B 的 DeepSeek 4.1 Flash；对比去年十二月只有 9 分的 Mistral Large 3 是巨大进步——虽不到 Fable 级，但 Mistral 重新回到距前沿约半年的梯队。
推荐理由
欧洲开源旗舰重新回到牌桌的量化证据：万亿参数+自建 Blackwell 集群+AA 位次跳升，选型开放权重模型的团队值得对照。
主体与标签
Mistral#模型发布#开源/仓库
正文
Introducing Mistral Large 4: Le chonk
The preview is available via their API. They promise to release the open weights model at the "end of this month".
The model only supports two reasoning levels - "none" and "high" - via the Mistral API. Here are both pelicans - the "high" one looks better, though surprisingly it only used 2,717 output tokens compared to "none" which used 3,275:
On Artificial Analysis it scores 38 , just behind DeepSeek 4.1 Flash, which is a 552B model. It's a huge improvement on last December's Mistral Large 3, which drew this terrible pelican and scored 9 on AA .
It's certainly not a Fable-class model, but it's great to see Mistral put out a model that's back to being maybe about 6 months behind the frontier. Via Hacker News
Tags: ai , generative-ai , llms , mistral , pelican-riding-a-bicycle , llm-release