Model release · MiniMax
MiniMax M3.1 Flash
- Model released
- Review updated
MiniMax's efficient everyday workhorse model; the stealth model Space Bunny, which topped several leaderboards, was an earlier version. MiniMax reports 62% on Automation-Bench, 30% on ALE, 74% on DeepSWE and 91% on BabyVision, at one third the cost of M3 and three times the speed.
Availability
MiniMax API at a flat $0.15 per 1M input tokens and $0.60 per 1M output tokens, with cache reads at $0.015 per 1M; also offered through M Plan and the MiniMax Code desktop app. Model weights are due on October 23, 2026.