MiniMax Releases M3.1 Flash With Higher Benchmark Scores and Lower Task Costs Than M3

MiniMax released M3.1 Flash, identified by QbitAI as the model previously tested anonymously as Space Bunny on OpenRouter. The report cites 29.6% on Agents’ Last Exam, 73.7% on DeepSWE v1.1, and output speeds reaching 200+ TPS. Pricing per million tokens is 1 yuan for input, 4 yuan for output, and 0.1 yuan for cache reads, with reported overall usage costs 70% below M3.

Latent Space’s AINews recap covers DeepSeek V4.1 Flash 1017, MiniMax M3.1 Flash, and Mercury 3, while noting they are not yet expected to see very wide use. It reports that DeepSeek’s model has a 1M-token context window and scores 45 on the Artificial Analysis Intelligence Index, tying GLM-5.3 for second among open-weight models. The recap also includes benchmark results, pricing, availability, and caveats for the releases.

MiniMax released M3.1 Flash, a reasoning model scoring 42 on the Artificial Analysis Intelligence Index, up from M3’s 29, at less than half the cost per task. It supports text and images with up to one million tokens of context; API pricing is $0.15 per million input tokens and $0.60 per million output tokens. Artificial Analysis reports a 93 percent hallucination rate because it answers rather than declines uncertain questions; open weights are planned for October 23.

Artificial Analysis benchmarked MiniMax-M3.1-Flash at max effort, reporting an Intelligence Index score of 42, up 13 points from MiniMax-M3 and level with GLM-5.3-Flash, but behind DeepSeek V4.1 Flash 1017 at 45. Its cost per Index task was ~$0.22 versus MiniMax-M3's ~$0.51, using ~189M output tokens. On AA-Omniscience, it answered 56% correctly but attempted almost every question, yielding a 93% hallucination rate and an overall score of 16.