Hacker News· Philpax·· 3 小时前AI 评分63
Mistral 发布 Large 4：开放权重多模态 MoE 模型
Mistral Large 4
AI 导读
Mistral 发布 Large 4 开放权重多模态模型，采用 Mixture-of-Experts 架构。该模型具备 49B 激活参数与 1.05T 总参数，并配有 1.6B 视觉编码器。文档链接：https://docs.mistral.ai/models/mistral-large-4-0
正文
Mistral Large 4 is a state-of-the-art, open-weight, general-purpose multimodal model with a granular Mixture-of-Experts architecture. It features 49B active parameters and 1.05T total parameters, and a 1.6B vision encoder.
来源：Hacker News · docs.mistral.ai