Models · Updated 19 Oct, 11:52 pm IST
MiniMax releases M3.1 Flash model with big performance and cost gains
Why it matters for readers: If you're curious about AI, this shows how models are becoming cheaper and more capable, while still making many confident mistakes.
- M3.1 Flash scores 42 on the Artificial Analysis Intelligence Index, up from 29 for M3.1
- The model accepts text and images, handles up to one million tokens of context, and runs three times faster than M3.1
- MiniMax reports a cost per task of about $0.22, achieved by using far fewer output tokens than rivals.1
- On reasoning and factual benchmarks, M3.1 Flash outperforms GLM-5.3-Flash and Gemini 3.8 Flash on several tests but trails Deepseek V4.1 Flash 1017 on the overall index.1
- Artificial Analysis notes a 93% hallucination rate for M3.1 Flash because it attempts answers instead of declining when uncertain.1
Get a brief like this every morning
Uzha reads hundreds of sources and gives you the stories that matter for your work, with every source linked. Free.
Get started