MiMo V2.6 Pro is Xiaomi’s flagship omni-modal reasoning model for complex projects, long-horizon tasks, high-stakes work, cybersecurity, and research. Xiaomi describes it as a trillion-parameter model with full-modality support and top-tier agent capabilities.
The model page lists capabilities including omni-modal understanding, deep thinking, tool call, streaming, web search, structured output, and context caching. Xiaomi also says the model combines 3D spatial reasoning, multimodal perception, and computer-use capabilities.
| Developer | Xiaomi |
|---|---|
| Model ID | xiaomi/mimo-v2.6-pro |
| Release Date | September 21, 2026 |
| Context Window | 1,048,576 tokens (~1.0M) |
| Max Output Tokens | 131,072 tokens |
| Modalities | Image, Text, Video → Text |
| Capabilities | reasoning, streaming, structured (JSON schema) output, tool/function calling, vision |
Omni-modal input
Xiaomi states that MiMo V2.6 Pro jointly understands images, video, audio, and text. Third-party coverage also describes it as natively supporting text, image, video, and audio inputs.
3D and interactive world capabilities
MiMo V2.6 can decompose text, image, or video prompts into multiple tasks, coordinate multiple agents, and build interactive 3D scenes with visual verification. Xiaomi also says it can generate 3D objects and scenes in Blender from text or reference images.
Computer-use and embodied control
The model is described as having computer-use agent capabilities and as being able to control a Franka Panda robotic arm in embodied simulation using multi-view camera input and visual feedback.
Tool and structured-output support
Xiaomi’s model page explicitly lists tool call, streaming, web search, structured output, and context caching among the model’s features. OpenRouter likewise reports tool calling and structured outputs support for this model.
Creative production workflows
Xiaomi says MiMo V2.6 Pro can handle video creation end to end, including visual design, shot and motion sequencing, music composition, and beat-synced editing. It can also generate frontend interfaces or slide decks with structured layouts, interactive elements, and animation.
Interactive 3D content creation
Use MiMo V2.6 Pro for turning text, images, or video into playable 3D worlds, interactive scenes, and Blender assets for animation, 3D printing, or game development.
Agentic software and design workflows
Xiaomi positions the model for complex programming, agent tasks, web pages, slides, frontends, and design work that may involve Figma or image/video generation tools.
Embodied simulation and robotics
The model is described as able to read multi-view camera feeds, reason continuously, and control a Franka Panda robotic arm to grasp, match colors, and place objects precisely.
Video and educational production
Xiaomi says the model can produce creative and product promo videos, and also explain abstract concepts with coherent animations and narration aligned to visuals.
MiMo V2.6 Flash
Xiaomi presents Flash as the other main model in the MiMo V2.6 family and says it emphasizes a balance between performance, efficiency, and cost, while Pro is the flagship reasoning model.
Sources: https://mimo.xiaomi.com/mimo-v2-6 · https://aiidelist.com/blog/what-is-mimo-v2-6 · https://ai-tldr.dev/models/mimo-v2-6-pro/
Measured on the AI/ML API gateway across live traffic. Figures are medians unless stated. Last updated 2026-10-12.
MiMo V2.6 Pro is documented as suited for complex software engineering, long-running agent tasks, and professional workflows.
It is also described as a fit for cybersecurity and research, especially high-stakes work and complex projects.
Another documented use case is new materials R&D, including literature review, molecular design, and automated dry-lab experiments for PFAS capture.
It is also shown helping researchers with material design and computational screening.
import requests r = requests.post( "https://api.aimlapi.com/v1/chat/completions", headers={"Authorization": "Bearer " + AIMLAPI_KEY}, json={ "model": "xiaomi/mimo-v2.6-pro", "messages": [ { "role": "user", "content": "Hello!" } ] }, ) print(r.json())
const r = await fetch("https://api.aimlapi.com/v1/chat/completions", { method: "POST", headers: { Authorization: `Bearer ${process.env.AIMLAPI_KEY}`, "Content-Type": "application/json", }, body: JSON.stringify({ "model": "xiaomi/mimo-v2.6-pro", "messages": [ { "role": "user", "content": "Hello!" } ] }), }); console.log(await r.json());
curl -X POST https://api.aimlapi.com/v1/chat/completions \ -H "Authorization: Bearer $AIMLAPI_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"xiaomi/mimo-v2.6-pro","messages":[{"role":"user","content":"Hello!"}]}'
OpenAI-compatible — swap the base URL and it works with your existing SDK.
| Type | Price |
|---|---|
| Input | |
| Output | |
| Cached input |
Cached input is billed at roughly a hundredth of the fresh-input rate, which matters for agent loops that re-send the same long prefix every turn.
| Benchmark | Score | What it measures | Source | Retrieved |
|---|---|---|---|---|
| Intelligence | 46.3 | Composite score across standardised reasoning, knowledge and problem-solving evaluations, measured independently by Artificial Analysis | Source | September 22, 2026 |
| Model | Input | Output | Context | Best for |
|---|---|---|---|---|
MiMo V2.6 Pro This page | Agentic workflows and structured output | |||
| Reasoning + agents | ||||
| Reasoning + agents |
MiMo V2.6 Pro is Xiaomi’s flagship reasoning model, built for complex projects, long-horizon tasks, high-stakes work, cybersecurity, and research.
It supports text, images, video, and audio as input.
It produces text output.
Yes. Xiaomi lists streaming as one of the model’s supported capabilities.
Yes. Xiaomi lists tool call support for the model.
Yes. Xiaomi lists structured output as a supported capability.
Yes. The model supports joint understanding of images and video, and Xiaomi describes it as omni-modal.
Xiaomi says it can coordinate multiple agents, handle computer-use tasks, and build interactive 3D scenes from text, image, or video prompts.
Yes. Xiaomi says it can generate 3D objects and scenes in Blender, create frontends and slide decks, and assemble video and music workflows.
Yes. Xiaomi describes it as omni-modal, with joint understanding of text, images, video, and audio.
