MiniMax Launches M3.1 Flash Model, Cuts Agentic Task Costs by 70%

MiniMax released the M3.1 Flash model on October 19, 2026, as the latest addition to its M3 series. The company says the model cuts costs on agentic tasks by about 70% compared to the previous generation and runs 3x faster than M3. API pricing is set at RMB 1 per million input tokens, RMB 4 per million output tokens, and RMB 0.1 per million cache reads, with model weights opening on October 23.

M3.1 Flash scores 62% on Automation-Bench and 30% on ALE in the Cowork category, 74% on DeepSWE for coding, and 91% on BabyVision for visual reasoning. MiniMax says the model delivers frontier-level intelligence at one-third the cost of M3. Its earlier version, "Space Bunny," topped global leaderboards with over 200T tokens. The API is available at platform.minimax.io, and users can also access MiniMax M Plan and the desktop MiniMax Code.