OpenCraft

xiaomi · chat model

Try MiMo-V2.6-Pro onlinenew

xiaomi/mimo-v2.6-pro

Use MiMo-V2.6-Pro in your browser. Explore pricing and capabilities below, or start a chat in OpenCraft.

Input price$0.435per M tokenscheaper than 45% of chat models
Output price$0.87per M tokenscheaper than 65% of chat models
Cache read$0.0036per M tokens
Context window1Mtokensof the largest window in the catalog (10M)

MiMo-V2.6-Pro pricing: what a turn costs

Estimates from the published token rates above — real turns vary with tool calls and transcript length.

  • Quick answer2K tokens in, 600 out$0.0014
  • 10-slide deck build60K tokens in, 9K out$0.034
  • 150-page PDF read200K tokens in, 4K out$0.09

Specs and listed prices from OpenRouter. Actual usage cost depends on the serving provider.

How to use MiMo-V2.6-Pro

  1. Select “Try MiMo-V2.6-Pro online” to open OpenCraft with this model selected.
  2. Sign in or create an account if prompted.
  3. Type a question or describe a task, then send your message.

To compare answers, start a separate chat with another model and use the same prompt.

MiMo-V2.6-Pro capabilities

tool-usevisionreasoning

text, image, video, audio in → text out

What is MiMo-V2.6-Pro?

MiMo-V2.6-Pro is Xiaomi’s flagship open-weights model, released September 22, 2026 (China time) under the MIT license. It is a sparse mixture-of-experts model with 1.02 trillion total parameters, 42 billion active per token, and a one-million-token context window. It takes text, image, audio and video input and returns text, with reasoning and tool calls.

Is it the best open-source model?

At launch it scored 46 on the Artificial Analysis Intelligence Index, the highest of any open-weights model on that index, ahead of GLM-5.3 and Kimi K3. That is one aggregate benchmark, not an OpenCraft measurement; try your own task before switching.

How much does MiMo-V2.6-Pro cost?

OpenRouter lists it at $0.435 per million input tokens and $0.87 output, close to the price of small models despite its size. MiMo-V2.6-Pro-UltraSpeed runs the same model up to 20× faster at ten times the price. Serving provider, reasoning and tool calls affect what a real turn costs.

MiMo-V2.6-Pro vs Flash

MiMo-V2.6-Flash is the smaller sibling (309 billion total parameters, 15 billion active) at $0.14 input and $0.28 output, with the same inputs and context window. Compare Flash and Pro, V2.5 Pro and V2.6 Pro, or MiMo-V2.6-Pro and DeepSeek V4.1 Flash.

Can I run it myself?

Yes. Xiaomi published the Pro and Flash weights on Hugging Face, along with a technical report and its reinforcement-learning code. To use it without hosting anything, try it in OpenCraft.

Release facts checked September 22, 2026 against Xiaomi’s Hugging Face release, SiliconANGLE’s launch report and OpenRouter’s listing. For current rates and availability, use the live provider listings.

Create with OpenCraft

See example outputs and find a workflow for your next project.