MiniMax Research · 2026-06-01精选模型评分79
MiniMax 发布 M3 模型：1M 上下文、原生多模态与前沿编码能力
MiniMax M3: Frontier Coding, 1M Context, Native Multimodality — All in One Model2026-06-01Learn More >
导读
MiniMax 发布 M3 模型，采用自研稀疏注意力架构 MSA（MiniMax Sparse Attention），支持最高 1M token 上下文窗口，并原生支持图像和视频输入、可操作桌面电脑。
推荐理由
官方给出 M3 在编码、智能体与多模态三项能力上的基准分数和真实任务演示，可据此判断开源权重模型与闭源前沿的差距。
分类置信度
高
- 官方或准官方来源，且无冲突类别。
- 正文、可信日期、主分类和证据摘录完整。
来源级别：官方一手
事件类型：研究发布
证据摘录
- “MiniMax M3 is officially released today.”
- “You can experience MiniMax M3 right away through MiniMax Code, the Token Plan, and our API services.”
- “The priority channel is currently enabled through sales support and is expected to open to all users in a few days.”
- “Over the next 10 days, we will release the model’s technical report and open-source the corresponding model weights.”
- “calls with ≤512K input tokens are billed at the standard rate, covering the vast majority of conversation and coding scenarios, while calls above 512K are billed at a higher long-context rate, mainly intended for high-load scenarios such as ultra-long document parsing and full-repository code understanding.”
来源：MiniMax Research · minimax.io