On October 23, MiniMax announced the open sourcing of the MiniMax M3.1 Flash model weights, which have a total of 428 billion parameters and 23 billion active parameters. Compared to the previous generation M3 model, the M3.1 Flash has achieved a comprehensive enhancement in model capabilities. According to official data, within 72 hours of its launch, the number of tokens processed by real users exceeded 100 trillion. On a technical level, M3.1 Flash improves inference efficiency through optimizations such as full model sparse attention, quantization-aware training (QAT), and power-aware scheduling. In terms of ecosystem, MiniMax collaborates with chip manufacturers including Huawei Ascend, T-head, Kunlun Core, Moore Threads, Biren, Haiguang, Muxi, Tian Shu, NVIDIA, and AMD, as well as cloud platforms like Together AI, Fireworks, Nebius, Baseten, Modal, and DigitalOcean to facilitate access on the first day of open source. Communities such as vLLM, SGLang, Zhiyuan FlagOS, and platforms like Hugging Face and Modao also support the deployment of the model across different chips and development environments.