Open models

MiniMax publishes M3.1 Flash weights

MiniMax_AI announced public weights for MiniMax M3.1 Flash, describing a model with 428B total parameters and 23B activated per token. Its reported results include 73.7% on DeepSWE v1.1, 61.7 on OSWorld V2 and 82.2 on CyberGym.

MiniMax_AI said Sparse Attention and DSpark speculative decoding improve inference efficiency. vllm_project announced launch support in vLLM, including native NVFP4 KV, NVFP4 MoE weights and DSpark speculative decoding in the serving stack.