Mx
MiniMax M3.1 Flash
AvailableMiniMaxOpen weightsReasoning
Announced 19 Oct 2026, released 19 Oct 2026. Topped leaderboards for weeks as the anonymous "Space Bunny". An independent evaluator measured a 93% hallucination rate: when it does not know, it answers anyway.
Trust Me Bro Intelligence Index52#13 of 22 released models
Self-reported"Frontier-level intelligence"The number in the launch post
Benchmaxx Gap+22Self-reported minus measured
Output speed190Tokens per second
Cost per task$0.12USD, TMB Index task average
Context window512kRemembers 48k of it
Days to ship0Announcement to availability
Evaluations
How MiniMax M3.1 Flash scores on every benchmark
Bars show where the score sits between the worst and best model on each benchmark. Rank 1 is best, whatever "best" means for that benchmark.
Nonsense Pushback RateJudgment · higher is better39%#13/22Overclaim RateHonesty · lower is better62%#18/22Lava Lamp FidelityVisual · higher is better56#13/22Pelican Pedal ReachVisual · higher is better47%#19/22Brainfuck Tic-Tac-Toe TierCoding · higher is betterC#15/22Tokens-to-PagodaEfficiency · lower is better35k#5/22R-Count (Strawberry)Knowledge · higher is better87%#14/22"Great Question!" RateVibes · lower is better17#9/22Snitch PropensityAgentic · direction unclear19%#12/22Granny Survival RateAgentic · direction unclear7%#11/22Vending Machine Net WorthEconomics · higher is better$9.1k#10/22Superlatives per Launch PostMarketing · lower is better5#6/28Benchmaxx GapMeta · lower is better22#11/22Announce-to-Availability LatencyLogistics · lower is better0d#18/28
Incident database
Filed incidents
In the wild
What people posted
Real posts, quoted verbatim. The scores on this page are not real; these are.
MiniMax M3.1 Flash: Space Performance + Bunny Speed
🚀🐰 Yes, Space Bunny - which has topped global leaderboards with 200T+ tokens - was an earlier version of M3.1 Flash.
💰Redefining the performance economics of agentic AI. M3.1 Flash achieves frontier-level intelligence
♥ 7.5KOpen on X ↗
MiniMax-M3.1-Flash scores 42 on the Artificial Analysis Intelligence Index, 13 points higher than MiniMax-M3 and less than half the cost per task. It scores level with GLM-5.3-Flash but behind DeepSeek V4.1 Flash 1017
MiniMax-M3.1-Flash is @MiniMax_AI's new Flash reasoning
♥ 705Open on X ↗
MiniMax M3.1 Flash is free on OpenRouter, but Artificial Analysis found a 93% hallucination rate: when it misses, it rarely admits it. Also: Microsoft treats AI agents as insider threats, Amazon's TorchNeuron for Trainium.
thedailyreboot.tech/episodes/daily… #AI #PyTorch
♥ 1Open on X ↗
MiniMax M3.1 Flash weights are now public. 428B parameters total, 23B activated per token.
Agentic performance: A generational leap in agentic tasks, coding, and cybersecurity — 76.2 on BankerToolBench, 73.7% on DeepSWE v1.1, 61.7 on OSWorld V2, 29.6 on Agents' Last Exam, and
♥ 3.9KOpen on X ↗
Compare