all the models — AI benchmark observatory
← Models

MiniMax M3.1 Flash

MiniMax · proprietary · minimax-m3.1-flash

MiniMax M3.1 Flash is the flash successor in the MiniMax M3 family, released 2026-10-19. It is a multimodal reasoning model with a 1M-token context window, native image and video input, and text output. Thinking is always on (cannot be disabled) with effort levels low/medium/high/xhigh/max (default max). Described as open-weight with weights to be released later; no Hugging Face checkpoint is available yet. The anonymous "Space Bunny" model was an early version. Available via the MiniMax API (Anthropic-compatible and OpenAI-compatible endpoints), M Plan, and MiniMax Code.

BabyVisionCyberGymBankerToolBe

Benchmark scores

BenchmarkScore
BabyVision0.91
CyberGym0.82
BankerToolBench0.76

Pricing

  • MiniMax$0.15 / $0.60

Input / output per 1M tokens

AA metrics

Intelligence
—
Speed tok/s
116.19
Blended $/M
—

Arena Elo

  • No Arena snapshot linked.