MiniMax releases M3.1 Flash with stronger agent benchmarks and lower pricing

What happened

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.

Reported by 2 sources

The story so far

MiniMax released MiniMax-M3.1-Flash through its API, and Artificial Analysis measured a higher Intelligence Index score and lower cost per Index task than MiniMax-M3. At maximum effort, the new model scored 42 versus M3’s 29, tying GLM-5.3-Flash and trailing DeepSeek V4.1 Flash 1017 at 45. Artificial Analysis estimated costs at approximately $0.22 per Index task, compared with $0.51 for M3. The model supports text and image inputs with up to one million tokens of context. The Decoder reported API prices of $0.15 per million input tokens and $0.60 per million output tokens. QbitAI reported prices of 1 yuan per million input tokens, 4 yuan per million output tokens and 0.1 yuan per million cache-read tokens, and cited overall usage costs 70% below M3. Its weights remain closed, with MiniMax planning an open-weights release on October 23. Artificial Analysis also found a weakness on AA-Omniscience: the model answered 56% of questions correctly but attempted almost every question, yielding a reported hallucination rate of 93% and an overall score of 16 on that evaluation.

Keep reading

Trending now

1. 1Mistral Releases Voxtral 3 Speech Models, Voice Agents API and LiveKit-Based SDK
2. 2DeepSeek Open-Sources DeepInfer and API Components, With Model and Hardware Limits
3. 3OpenAI Acquires Orchestra Bio to Expand AI Tools for Drug Development
4. 4Google Rolls Out Gemini Personal Intelligence to Android Auto, With Google Built-In Cars to Follow