ModelScope 汇总 M3.1 Flash 规格与跑分：Automation-Bench 62.8、DeepSWE
ModelScope 列出 M3.1 Flash 的 428B MoE / 23B 激活、1M 上下文、原生图像视频输入，以及 Automation-Bench 62.8、DeepSWE 73.7%、OSWorld 2.0 61.7、CyberGym 82.2 等官方跑分，并给出 SGLang/vLLM 自托管指引。
1M tokens, native image and video input, and 23B active parameters per token. MiniMax-M3.1-Flash has landed as a 428B MoE for coding, computer use, and tool-driven agents. 🤖 https://modelscope.ai/models/MiniMax/MiniMax-M3.1-Flash 📄 https://modelscope.ai/papers/2606.13392 🏆 Top result for agent execution: Scores 62.8 on Automation-Bench v1.0.6, the highest result in the model card’s reported comparison. 💻 Strong at coding and computer use: Reaches 73.7% on DeepSWE v1.1, 61.7 on OSWorld 2.0, and 29.6 on Agents’ Last Exam. 🔐 Built for cybersecurity work: Scores 82.2 on CyberGym for vulnerability discovery and validation. ⚡ Efficiency is built in: Sparse attention throughout, QAT NVFP4, and the included DSpark draft head support efficient inference and speculative decoding. 🛠️ Ready to self-host: Model weights are available with SGLang and vLLM deployment guidance.