MiniMaxAI/MiniMax-M3.1-Flash

MiniMax has released the weights of MiniMax-M3.1-Flash, launched via API on 19 October. It is a 428B-parameter multimodal mixture-of-experts model (23B active per token) that reads text, images and video, outputs text and handles up to 1 million tokens of context. It targets coding, agents, computer use and cybersecurity; MiniMax says it is 3× faster and 3× cheaper than M3. Official API price: $0.15 / $0.60 per million input / output tokens. ⚠️ License: “MiniMax Community”, not a standard open-source licence; check it before commercial use.

See original publication↗MiniMax (Hugging Face officiel) · huggingface.co