Xiaomi · language model
MiMo-V2.6-Pro-UltraSpeed: pricing, context and benchmarks
MiMo-V2.6-Pro-UltraSpeed is a language model from Xiaomi released on September 21, 2026. It costs $4.35 per million input tokens and $8.7 per million output tokens, with a 1M-token context window. No independent score has been published for it so far.
| Price | 3/30 | |
|---|---|---|
| Context | 14/15 | |
| Features | 11/15 |
What you need to know about MiMo-V2.6-Pro-UltraSpeed
MiMo-V2.6-Pro-UltraSpeed was released by Xiaomi (China) on September 21, 2026. It accepts text, images, video, and audio as input and produces text. Its context window reaches 1,048,576 tokens, larger than 93 % of the models in the registry, and it can write up to 131,072 tokens in one go.
It is a recent model: real-world feedback and independent scores are still scarce.
MiMo-V2.6-Pro-UltraSpeed pricing
On OpenRouter, MiMo-V2.6-Pro-UltraSpeed costs $4.35 per million input tokens and $8.7 per million output tokens. Counting three tokens read for every token written, its average price is $5.438 per million, more expensive than 89 % of paid models (7.1 times the median of $0.76).
A concrete example: a task that sends 2,000 tokens (about 1,500 words) and receives a 500-token answer costs $0.0131. At 1,000 tasks a day, budget $392 a month. The price calculator redoes this with your own volumes.
Tokens already sent and cached are billed at $0.036 per million.
Performance and benchmarks
Epoch AI does not yet publish a capabilities index for MiMo-V2.6-Pro-UltraSpeed. We therefore show no made-up performance rating: only price, context and features count towards its score.
Features, privacy and local use
- It supports tool calling, so it can drive an AI agent, call APIs or MCP servers.
- It can answer in an enforced JSON format (structured outputs).
- It is a reasoning model: it "thinks" before answering, which helps on hard tasks but uses more output tokens.
Its weights are not published: it can only be used through the Xiaomi API or a broker such as OpenRouter, which means sending your data off your machines.
Other Xiaomi models
| Model | Released | Avg. price | ECI |
|---|---|---|---|
| MiMo-V2.6-Flash | Sep 21, 2026 | $0.18 | — |
| MiMo-V2.6-Pro | Sep 21, 2026 | $0.54 | — |
| MiMo-V2.5-Pro | Apr 22, 2026 | $0.38 | — |
| MiMo-V2.5 | Apr 22, 2026 | $0.15 | — |
Verdict
MiMo-V2.6-Pro-UltraSpeed scores 47/100 on our grid, which only uses published figures.
Strengths
- Very large context (1M tokens): long documents or a whole codebase fit in one request.
- Agent-ready: tool calling and structured JSON output.
Limitations
- Expensive: costlier than 89 % of paid models.
- No independent score published: performance cannot be verified.
Avoid if…
- you process large volumes on a tight budget;
- your data must not leave your servers;
- your tasks are mostly simple: a non-reasoning model will cost less.
Alternatives
Comparisons and tools
Learn more
Frequently asked questions
How much does MiMo-V2.6-Pro-UltraSpeed cost?
$4.35 per million input tokens and $8.7 per million output tokens on OpenRouter (checked October 25, 2026). A task with 2,000 tokens read and 500 written costs $0.0131.
How big is MiMo-V2.6-Pro-UltraSpeed's context window?
1,048,576 tokens, roughly 786,000 words, with a maximum of 131,072 output tokens.
Can MiMo-V2.6-Pro-UltraSpeed run locally?
No, Xiaomi does not publish its weights: it is API-only.
Is MiMo-V2.6-Pro-UltraSpeed suitable for an AI agent?
Technically yes: it supports tool calling and structured outputs. Then check its cost for your volume.
What are the alternatives to MiMo-V2.6-Pro-UltraSpeed?
The closest models on price and performance are Gemini 3.1 Pro Preview, Gemini 3.1 Pro Preview Custom Tools, and Ember-1.
Related articles
- Mistral AI: models, pricing and agentsOctober 25, 2026
- LLM benchmarks: how to read model rankingsOctober 24, 2026
- Running a local LLM: hardware, software, and limitationsOctober 24, 2026
- Open-source or closed models for your AI agents?October 5, 2026
Neutrality. No affiliate links. No vendor pays to be reviewed, rated or ranked. Reviews are directed by François Chollet, neutral and factual, and every data point is dated and sourced. How we score