MiniMax M3.1 Flash

Use it to build coding and agentic workflows, process image or video inputs, and perform computer-use tasks.

Availability

Hugging Face model page and API documentation linked

Pricing model

$0.15 input, $0.60 output, and $0.015 cached input per 1M tokens

Deployment

API access is linked; model is hosted on Hugging Face.

Privacy

Not established

Licence

Not established

• M3.1 Flash is available through the MiniMax API, M Plan, and MiniMax Code. ^[1]

  M3.1 Flash is now available through the MiniMax API, M Plan, and MiniMax Code.

• Pricing per 1M tokens is $0.15 input, $0.60 output, and $0.015 cached input. ^[1]

  Flat pricing per 1M tokens: $0.15 input · $0.60 output · $0.015 cached input.

• MiniMax says online inference is nearly three times faster and usage costs are 70% lower than its predecessor. ^[1]

  As a next-generation Flash model, M3.1 Flash delivers a generational leap in agentic tasks, coding, Computer Use and cybersecurity, while making online inference nearly three times faster and reducing usage costs by 70%.

• The model accepts text, image, and video inputs and produces text output. ^[2]

  It supports text, image, and video inputs with a context window of up to **1M tokens**.

• The context window supports up to 1,000,000 tokens. ^[2]

  It supports text, image, and video inputs with a context window of up to **1M tokens**.

• The model has approximately 428B total parameters, with 23B activated per token. ^[2]

  - **Parameters:** Approximately 428B total; 23B activated per token

• The model uses MiniMax Sparse Attention throughout and includes a DSpark speculative-decoding draft head. ^[2]

  Building on MiniMax-M3, M3.1-Flash improves task performance and inference efficiency with sparse attention throughout the model, quantization-aware training, and DSpark speculative decoding.

• The announcement links to the model on Hugging Face and to MiniMax API text-generation documentation.