Home/AI agents/MiniMax M3.1 Flash
MiniMax M3.1 Flash
MiniMax multimodal model for text, image, and video input, with a 1 million token context.
MiniMax says it officially rolled M3.1 Flash out on October 19, 2026. It is a native multimodal mixture-of-experts model with about 428 billion total parameters and 23 billion active. It takes text, image, and video input and returns text. Context is 1 million tokens. Thinking depth is tunable (effort: low, medium, high, xhigh, max; default max). Weights are on Hugging Face as MiniMaxAI/MiniMax-M3.1-Flash under the minimax-community license, which is not an OSI license. It is available through the MiniMax API (OpenAI- and Anthropic-compatible, model id MiniMax-M3.1-Flash), the M Plan, and MiniMax Code. Lab-reported: 73.7% DeepSWE v1.1 and 61.7 on OSWorld V2.
Where it ranks
16 in AI agents
A multimodal model API with a 1 million token context.
See the ai agents list17 in Vibe coding
A coding-capable model you call from MiniMax Code or the API.
See the vibe coding list
Pricing
Paid
MiniMax lists API prices per 1 million tokens: $0.15 input, $0.60 output, and $0.015 cached input. Source: platform.minimax.io pricing page. Check MiniMax for M Plan and MiniMax Code prices.
Checked
Added October 23, 2026
Last verified October 23, 2026
Dated releases
MiniMax rolled out M3.1 Flash, a multimodal model with about 428 billion total parameters and a 1 million token context.