AI 导读
MiniMax 发布 MiniMax M3，称其为首个结合三项前沿能力的开源权重模型。官方给出的编码与智能体成绩包括 59.0% SWE-Bench Pro、66.0% Terminal Bench 2.1 和 74.2% MCP Atlas。模型用 MiniMax Sparse Attention 将上下文扩展至 1M，并原生支持多模态，权重与技术报告约 10 天后发布。
正文
Minimax Team has texted us at
- 8 PM on a Sunday
- 11 AM on a Wednesday
I genuinely do not know which timezone these people they work in and that's how I know they will succeed.
Kudos for the Minimax 3 release
Introducing MiniMax M3: The First Open-Weights Model to Combine Three Frontier Capabilities - Coding & Agentic Frontier: 59.0% SWE-Bench Pro, 66.0% Terminal Bench 2.1, 34.8% SWE-fficiency, 28.8% KernelBench Hard, 74.2% MCP Atlas - MiniMax Sparse Attention scales context to 1M - Natively Multimodal from Step Zero API: platform.minimax.io Token Plan: platform.minimax.io/subscrib… 🚀New! MiniMax Code: code.minimax.io Weights & Tech Report in ~10 Days在 X 查看被引用的帖子
来源：@arafatkatze · x.com