Xiaomi
MiMo V2.6 Pro
Xiaomi's flagship for long agent runs and software engineering. Near Claude Opus 5 on coding agents at a fraction of the price.
Replaces MiMo V2.5 Pro: same input and output price, context 1.1M → 1M, max output 131K → 131K.
Run this model
xiaomi/mimo-v2.6-pro
Use an endpoint supported by this model.
Get a key for this setup →const response = await fetch('https://api.minirouter.sh/v1/chat/completions', { method: 'POST', headers: { Authorization: `Bearer ${process.env.MINIROUTER_KEY}`, 'Content-Type': 'application/json', }, body: JSON.stringify({ model: 'xiaomi/mimo-v2.6-pro', messages: [{ role: 'user', content: 'Why is the sky blue?' }], max_tokens: 1024, }),}) const data = await response.json()console.log(data.choices[0].message.content)Use Chat Completions with an explicit max_tokens. Text and image input are enabled; audio and video are not yet.
Model overview
- Input / 1M
- $0.45675
- All-in MiniRouter rate
- Output / 1M
- $0.9135
- All-in MiniRouter rate
- Context
- 1M
- 1,048,576 tokens
- Max output
- 131K
- tokens
- DeepSWE v1.1
- 71.9%
- Up from 19.0% on V2.5 Pro
- Architecture
- 1.02T MoE
- 42B active, MIT weights
- Announced
- 22 Sept 2026
- publisher release
Publisher evidence
A generational jump on agent work.
Xiaomi's own harness and settings. Compare within this table, not against other publishers' numbers.
Xiaomi-reported
Coding, agents and tools
- MiMo V2.6 Pro
- Comparison models
Coding agent
DeepSWE v1.1
- MiMo V2.6 Pro71.9%
- MiMo V2.6 Flash67.9%
- MiMo V2.5 Pro19.0%
- Claude Opus 574.0%
- GPT-5.6 Sol73.0%
0 to 100%
General agent
AutomationBench v1.0.6
- MiMo V2.6 Pro53.1%
- MiMo V2.6 Flash52.3%
- MiMo V2.5 Pro16.0%
- Claude Opus 550.3%
- GPT-5.6 Sol45.8%
0 to 100%
Tool use
Toolathlon-Verified
- MiMo V2.6 Pro76.9%
- MiMo V2.6 Flash73.6%
- MiMo V2.5 Pro49.1%
- Claude Opus 580.6%
- GPT-5.6 Sol74.9%
0 to 100%
Terminal
Terminal Bench 2.1
- MiMo V2.6 Pro89.9%
- MiMo V2.6 Flash87.6%
- MiMo V2.5 Pro65.2%
- Claude Opus 589.1%
- GPT-5.6 Sol88.8%
0 to 100%
View chart data
| Benchmark | MiMo V2.6 Pro | MiMo V2.6 Flash | MiMo V2.5 Pro | Claude Opus 5 | GPT-5.6 Sol |
|---|---|---|---|---|---|
| DeepSWE v1.1 | 71.9% | 67.9% | 19.0% | 74.0% | 73.0% |
| AutomationBench v1.0.6 | 53.1% | 52.3% | 16.0% | 50.3% | 45.8% |
| Toolathlon-Verified | 76.9% | 73.6% | 49.1% | 80.6% | 74.9% |
| Terminal Bench 2.1 | 89.9% | 87.6% | 65.2% | 89.1% | 88.8% |
Pick a variant
One family, three price points.
Flash trades a few points for a third of the price. UltraSpeed runs Pro faster for ten times the price.
Xiaomi release note- 01
xiaomi/mimo-v2.6-pro
Default for coding agents and long tasks.
- 02
xiaomi/mimo-v2.6-flash
Within a few points of Pro on most agent tests.
- 03
xiaomi/mimo-v2.6-pro-ultraspeed
Pro quality, up to 20x faster, for latency-sensitive loops.
One route, standard clients.
Authenticate with a MiniRouter bearer key. Caller-provided upstream credentials are not used on this route.
Read model setup docs →https://api.minirouter.sh/v1xiaomi/mimo-v2.6-pro/v1/chat/completions/v1/messages/v1/responsesmax_tokens, temperature, stop, tools, tool_choice, reasoning, include_reasoning
Not enforced; a requested format is dropped and the answer is plain text.
Configured provider
Priority follows the live catalog order.
Input: text, image · Output: text
| Provider | Priority | Input / 1M | Output / 1M | Parameters | Access |
|---|---|---|---|---|---|
vercel | 1 | $0.45675 | $0.9135 | max_tokens, temperature, stop, tools, tool_choice, reasoning, include_reasoning | MiniRouter balance |