MiniMax 发布 M3:前沿编码、1M 上下文与原生多模态合一
MiniMax 正式发布 M3,采用自研 MSA 稀疏注意力架构,支持最高 1M token 上下文,并原生支持图像与视频输入、可操作桌面电脑。官方称 M3 是目前首个同时具备这三项能力的开源权重模型,在 SWE-Bench Pro 得分 59.0%、Terminal-Bench 2.1 得分 66.0%、MCP Atlas 得分 74.2%。M3 已通过 MiniMax Code、Token Plan 和 API 开放使用,官方表示将在未来 10 天内发布技术报告并开源模型权重。

model weights. Today, models are being updated at such a fast pace that it is easy to forget this is still a matter of steady, incremental progress. It follows its own objective laws, and it rewards teams that move forward solidly in accordance with those laws. Just as we believed at the very beginning of our founding, we will do our utmost to continuously improve the intelligence of our models and make them available to more users. Thank you for your trust, suggestions, and criticism. Intellig
匹配字段正文


