XiaomiLanguage modelActive

MiMo V2.6 Pro UltraSpeed

MiMo V2.6 Pro UltraSpeed has no overall index value yet because there are too few independent benchmark results.

Intelligencenot ranked yet

n/aOverall indexhigher is better

Speednot measured

n/ano measurement yet

PriceRank 79 of 95

Input $4.35 · Output $8.70

$5.44per 1M tokenslower is better

Contextn/a

n/atokens at once

PopularityRank 673 of 876

0out of 100same as last month

Summary

It is expensive: $5.44 per 1M tokens (input and output blended); a typical model costs $1.56.

MiMo V2.6 Pro UltraSpeed is Xiaomi’s accelerated API and desktop mode for MiMo V2.6 Pro. Xiaomi refers to the regular Pro checkpoint for the released weights. The name describes lower-latency serving within the V2.6 family.

Xiaomi reports acceleration of up to 20 times for UltraSpeed. That is a vendor figure and depends on the platform and workflow.

See for yourself: one prompt, one attempt

SVG short test

One SVG task with the same fixed settings for every model, without hidden thinking.

MiMo V2.6 Pro UltraSpeed

Earlier tests

The response reached the token limit. No complete recording is available.

Pelican on a bicycle

Prompt

Generate an SVG of a pelican riding a bicycle

Oct 6, 2026 · 16 s · $0.07

Not yet visually reviewed. A successful recording does not confirm correct physics or full compliance with the prompt.

Test details

Run date: Oct 6, 2026
API provider: openrouter
Requested API model: xiaomi/mimo-v2.6-pro-ultraspeed
Reasoning setting: high
Token limit including reasoning: 8,192 tokens
Output tokens: 8,192 tokens
Cost of the attempt: $0.0713
Response time: 15.9 s
API model reported in response: xiaomi/mimo-v2.6-pro-ultraspeed
Pinned endpoint: Xiaomi | xiaomi/mimo-v2.6-pro-ultraspeed-20260921
Temperature: Provider default
Test protocol: community-visual-tests-v1
Technical checks: Failed

Earlier attempts (2)

Oct 2, 2026API request failedReasoning setting: high
Oct 1, 2026API request failedReasoning setting: high

Task origin: Simon Willison

All benchmark results

Every published result with its source, test setup and comparison models.

Show all 29 published benchmark results

Measurements without matching peer values

These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.

DataLLM Lab Python function suite hidden-assert tasks passed (Show measurement, test conditions, and source)

Source value: 9
Score: 9
Metric: tasks passed
Unit: count
Benchmark version: DataLLM Lab nine-task Python function suite, run 2026-10-02
Category: coding
Direction: Higher is better
Sample: 9
Source type: independent-evaluator
Evaluator: DataLLM Lab executed API benchmark
Status: active
Retrieved at: 2026-10-16
Methodology: Independent evaluation. Exact endpoint xiaomi/mimo-v2.6-pro-ultraspeed through OpenRouter. Nine self-contained Python function tasks; generated code executed against hidden asserts in isolated python3 -I with a 12-second timeout. Temperature 0, max_tokens 4000, one scored attempt per task. All nine passed.
Context: A tiny, single-attempt micro-suite at ceiling performance. It does not distinguish model quality, measure multimodal or agentic work, or support generalized coding comparisons.

Source DataLLM Lab executed API benchmark

DataLLM Lab Python function suite mean per-call latency (Show measurement, test conditions, and source)

Source value: 2.5
Score: 2.5
Metric: arithmetic mean latency per call
Unit: seconds
Benchmark version: DataLLM Lab nine-task Python function suite, run 2026-10-02
Category: coding
Direction: Lower is better
Sample: 9
Source type: independent-evaluator
Evaluator: DataLLM Lab executed API benchmark
Status: active
Retrieved at: 2026-10-16
Methodology: Independent evaluation. Exact endpoint xiaomi/mimo-v2.6-pro-ultraspeed through OpenRouter. Nine calls in the same nine-task Python function suite; reported mean call latency 2.5 seconds. This is wall-clock latency, not token throughput.
Context: Single sweep and only nine calls; the value is not a stable service-level or general throughput estimate.

Source DataLLM Lab executed API benchmark

ToneBench Writing Score (Show measurement, test conditions, and source)

Source value: 80.6
Score: 80.6
Metric: composite score out of 100
Unit: points
Benchmark version: ToneBench October 2026 board, generated 2026-10-13T20:08:00Z
Category: English long-form writing
Direction: Higher is better
Sample: 50
Source type: independent-evaluator
Evaluator: Towards AI ToneBench
Status: superseded
Retrieved at: 2026-10-17
Methodology: Independent evaluation. Via OpenRouter, exact route xiaomi/mimo-v2.6-pro-ultraspeed; provider-default sampling, the provider applies thinking on; five independent drafts for each of ten English script briefs (50 drafts), run date 2026-10-08; four blind judges; closed-book with no tools.
Context: Specialized English long-form YouTube script writing in one creator's voice, not a general capability measure. Four blind judges score drafts; Length Adherence is computed and half of Slop Score is computed. The exact rubric is private. Published ± values for metrics are draft-to-draft standard deviations, not confidence intervals.

Source Towards AI ToneBench

ToneBench Writing Elo (Show measurement, test conditions, and source)

Source value: 1990
Score: 1990
Metric: board-relative pairwise writing rating
Unit: elo
Benchmark version: ToneBench October 2026 board, generated 2026-10-13T20:08:00Z
Category: English long-form writing
Direction: Higher is better
Lower confidence bound: 1947
Upper confidence bound: 2037
Source type: independent-evaluator
Evaluator: Towards AI ToneBench
Status: superseded
Retrieved at: 2026-10-17
Methodology: Independent evaluation. Via OpenRouter, exact route xiaomi/mimo-v2.6-pro-ultraspeed; provider-default sampling, the provider applies thinking on; five independent drafts for each of ten English script briefs (50 drafts), run date 2026-10-08; four blind judges; closed-book with no tools.
Context: Specialized English long-form YouTube script writing in one creator's voice, not a general capability measure. Four blind judges score drafts; Length Adherence is computed and half of Slop Score is computed. The exact rubric is private. Published ± values for metrics are draft-to-draft standard deviations, not confidence intervals.

Source Towards AI ToneBench

ToneBench Tone & Voice Match (Show measurement, test conditions, and source)

Source value: 78.2
Score: 78.2
Metric: Tone & Voice Match
Unit: points
Benchmark version: ToneBench October 2026 board, generated 2026-10-13T20:08:00Z
Category: English long-form writing
Direction: Higher is better
Sample: 50
Source type: independent-evaluator
Evaluator: Towards AI ToneBench
Status: superseded
Retrieved at: 2026-10-17
Methodology: Independent evaluation. Via OpenRouter, exact route xiaomi/mimo-v2.6-pro-ultraspeed; provider-default sampling, the provider applies thinking on; five independent drafts for each of ten English script briefs (50 drafts), run date 2026-10-08; four blind judges; closed-book with no tools.
Context: Specialized English long-form YouTube script writing in one creator's voice, not a general capability measure. Four blind judges score drafts; Length Adherence is computed and half of Slop Score is computed. The exact rubric is private. Published ± values for metrics are draft-to-draft standard deviations, not confidence intervals.

Source Towards AI ToneBench

ToneBench Writing Craft & Clarity (Show measurement, test conditions, and source)

Source value: 78
Score: 78
Metric: Writing Craft & Clarity
Unit: points
Benchmark version: ToneBench October 2026 board, generated 2026-10-13T20:08:00Z
Category: English long-form writing
Direction: Higher is better
Sample: 50
Source type: independent-evaluator
Evaluator: Towards AI ToneBench
Status: superseded
Retrieved at: 2026-10-17
Methodology: Independent evaluation. Via OpenRouter, exact route xiaomi/mimo-v2.6-pro-ultraspeed; provider-default sampling, the provider applies thinking on; five independent drafts for each of ten English script briefs (50 drafts), run date 2026-10-08; four blind judges; closed-book with no tools.
Context: Specialized English long-form YouTube script writing in one creator's voice, not a general capability measure. Four blind judges score drafts; Length Adherence is computed and half of Slop Score is computed. The exact rubric is private. Published ± values for metrics are draft-to-draft standard deviations, not confidence intervals.

Source Towards AI ToneBench

ToneBench Substance, Accuracy & Value (Show measurement, test conditions, and source)

Source value: 82.7
Score: 82.7
Metric: Substance, Accuracy & Value
Unit: points
Benchmark version: ToneBench October 2026 board, generated 2026-10-13T20:08:00Z
Category: English long-form writing
Direction: Higher is better
Sample: 50
Source type: independent-evaluator
Evaluator: Towards AI ToneBench
Status: superseded
Retrieved at: 2026-10-17
Methodology: Independent evaluation. Via OpenRouter, exact route xiaomi/mimo-v2.6-pro-ultraspeed; provider-default sampling, the provider applies thinking on; five independent drafts for each of ten English script briefs (50 drafts), run date 2026-10-08; four blind judges; closed-book with no tools.
Context: Specialized English long-form YouTube script writing in one creator's voice, not a general capability measure. Four blind judges score drafts; Length Adherence is computed and half of Slop Score is computed. The exact rubric is private. Published ± values for metrics are draft-to-draft standard deviations, not confidence intervals.

Source Towards AI ToneBench

ToneBench Continuity & Emotion (Show measurement, test conditions, and source)

Source value: 80
Score: 80
Metric: Continuity & Emotion
Unit: points
Benchmark version: ToneBench October 2026 board, generated 2026-10-13T20:08:00Z
Category: English long-form writing
Direction: Higher is better
Sample: 50
Source type: independent-evaluator
Evaluator: Towards AI ToneBench
Status: superseded
Retrieved at: 2026-10-17
Methodology: Independent evaluation. Via OpenRouter, exact route xiaomi/mimo-v2.6-pro-ultraspeed; provider-default sampling, the provider applies thinking on; five independent drafts for each of ten English script briefs (50 drafts), run date 2026-10-08; four blind judges; closed-book with no tools.
Context: Specialized English long-form YouTube script writing in one creator's voice, not a general capability measure. Four blind judges score drafts; Length Adherence is computed and half of Slop Score is computed. The exact rubric is private. Published ± values for metrics are draft-to-draft standard deviations, not confidence intervals.

Source Towards AI ToneBench

ToneBench YouTube Best Practices (Show measurement, test conditions, and source)

Source value: 78.6
Score: 78.6
Metric: YouTube Best Practices
Unit: points
Benchmark version: ToneBench October 2026 board, generated 2026-10-13T20:08:00Z
Category: English long-form writing
Direction: Higher is better
Sample: 50
Source type: independent-evaluator
Evaluator: Towards AI ToneBench
Status: superseded
Retrieved at: 2026-10-17
Methodology: Independent evaluation. Via OpenRouter, exact route xiaomi/mimo-v2.6-pro-ultraspeed; provider-default sampling, the provider applies thinking on; five independent drafts for each of ten English script briefs (50 drafts), run date 2026-10-08; four blind judges; closed-book with no tools.
Context: Specialized English long-form YouTube script writing in one creator's voice, not a general capability measure. Four blind judges score drafts; Length Adherence is computed and half of Slop Score is computed. The exact rubric is private. Published ± values for metrics are draft-to-draft standard deviations, not confidence intervals.

Source Towards AI ToneBench

ToneBench Hook Strength (Show measurement, test conditions, and source)

Source value: 80.6
Score: 80.6
Metric: Hook Strength
Unit: points
Benchmark version: ToneBench October 2026 board, generated 2026-10-13T20:08:00Z
Category: English long-form writing
Direction: Higher is better
Sample: 50
Source type: independent-evaluator
Evaluator: Towards AI ToneBench
Status: superseded
Retrieved at: 2026-10-17
Methodology: Independent evaluation. Via OpenRouter, exact route xiaomi/mimo-v2.6-pro-ultraspeed; provider-default sampling, the provider applies thinking on; five independent drafts for each of ten English script briefs (50 drafts), run date 2026-10-08; four blind judges; closed-book with no tools.
Context: Specialized English long-form YouTube script writing in one creator's voice, not a general capability measure. Four blind judges score drafts; Length Adherence is computed and half of Slop Score is computed. The exact rubric is private. Published ± values for metrics are draft-to-draft standard deviations, not confidence intervals.

Source Towards AI ToneBench

ToneBench Length Adherence (Show measurement, test conditions, and source)

Source value: 89.4
Score: 89.4
Metric: Length Adherence
Unit: points
Benchmark version: ToneBench October 2026 board, generated 2026-10-13T20:08:00Z
Category: English long-form writing
Direction: Higher is better
Sample: 50
Source type: independent-evaluator
Evaluator: Towards AI ToneBench
Status: superseded
Retrieved at: 2026-10-17
Methodology: Independent evaluation. Via OpenRouter, exact route xiaomi/mimo-v2.6-pro-ultraspeed; provider-default sampling, the provider applies thinking on; five independent drafts for each of ten English script briefs (50 drafts), run date 2026-10-08; four blind judges; closed-book with no tools.
Context: Specialized English long-form YouTube script writing in one creator's voice, not a general capability measure. Four blind judges score drafts; Length Adherence is computed and half of Slop Score is computed. The exact rubric is private. Published ± values for metrics are draft-to-draft standard deviations, not confidence intervals.

Source Towards AI ToneBench

ToneBench Slop Score (Show measurement, test conditions, and source)

Source value: 83.8
Score: 83.8
Metric: Slop Score
Unit: points
Benchmark version: ToneBench October 2026 board, generated 2026-10-13T20:08:00Z
Category: English long-form writing
Direction: Higher is better
Sample: 50
Source type: independent-evaluator
Evaluator: Towards AI ToneBench
Status: superseded
Retrieved at: 2026-10-17
Methodology: Independent evaluation. Via OpenRouter, exact route xiaomi/mimo-v2.6-pro-ultraspeed; provider-default sampling, the provider applies thinking on; five independent drafts for each of ten English script briefs (50 drafts), run date 2026-10-08; four blind judges; closed-book with no tools.
Context: Specialized English long-form YouTube script writing in one creator's voice, not a general capability measure. Four blind judges score drafts; Length Adherence is computed and half of Slop Score is computed. The exact rubric is private. Published ± values for metrics are draft-to-draft standard deviations, not confidence intervals.

Source Towards AI ToneBench

Page 1 of 3

Popularity compared

How often the model is searched for, used and mentioned, compared with every model of its category.

Popularity index

The 100 most popular models of this category, 0 to 100, higher is more popular

20 of 101 models

GPT-573

DeepSeek V4.1 Flash 101769

Kimi K368

DiffusionGemma 26B A4B IT63

MiMo V2.6 Pro UltraSpeed0

DeepSeek V4.1 Flash 1017

DiffusionGemma 26B A4B IT

MiMo V2.6 Pro UltraSpeed

As of November 2026 · Source: gradually.ai popularity index

Over time

Weekly values from October 18, 2026 to November 6, 2026

MiMo V2.6 Pro UltraSpeedAgnes 3.0 FlashAion 3.5Aion 3.5 MiniEmber-1

What is gaining popularity right now, across models, vendors, harnesses and AI subscriptions, is collected in All AI trends.

Similar models

More models of the same kind as MiMo V2.6 Pro UltraSpeed, chosen by model line, developer, size, price and release date.

All specifications

Architecture

Model class: High-speed mode for MiMo V2.6 Pro; Source: Model class

Input and output

Speed: Up to 20× faster inference according to Xiaomi; Source: Speed

Watermarking and provenance

AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.

AI watermarking: Unknown; no source
Watermarking technique: Not known; no source

Licence and access

Access: MiMo API and MiMo Desktop; Source: Access
Weights: No separate weights published; Source: Weights

API

API model ID: mimo-v2.6-pro-ultraspeed; Source: API model ID

Further details

Batch API: Not available; Source: Batch API

Prices

API pricesSubscriptions

1 offer, prices in US dollars per 1M tokens, blended from 3 parts input and 1 part output.

Provider	Input	Output	Cache read	Blended
Xiaomidirect	$4.35	$8.70	$0.036	$5.44

Direct price according to the developer, checked Oct 25, 2026. Open the LLM price index

More price tiers and rules

Cache price: US$0.036 / 1M tokens, overseas real-time API, cache hitSource
Input price: US$4.35 / 1M tokens, overseas real-time API, cache missSource
Output price: US$8.70 / 1M tokens, overseas real-time APISource

No Xiaomi subscription includes this model. Through other providers from $10 a month (Command Code GOAT, with limited credits). 4 plans in Germany.

Plan	Monthly list price	Note
Command Code GOAT	$10	with limited credit
Command Code Pro	$20	with limited credit
Command Code Max 10×	$100	with limited credit
Command Code Max 20×	$200	with limited credit

All plans in the AI subscription comparison

Frequently asked questions about MiMo V2.6 Pro UltraSpeed

How much does MiMo V2.6 Pro UltraSpeed cost?

Through the API $4.35 per 1M input tokens and $8.70 output. That is expensive.

PriceDollars per 1M tokens, input and output blended · lower is better

GPT-5.6 Terra$4.50

Gemini 3.1 Pro Preview$4.50

MiMo V2.6 Pro UltraSpeedRank 79 of 95$5.44

GPT-5.4$5.63

Kimi K3$6

Claude Sonnet 4.6$6

Claude Sonnet 4.5$6

Where can I use MiMo V2.6 Pro UltraSpeed?

Not in a Xiaomi subscription. Including other providers that makes 4 plans in Germany, plus the API.

Subscription comparison, prices in Germany incl. VAT

At Xiaomi

No Xiaomi subscription includes this model.

Through other providers

4 plans from 1 provider, from $10 a month (Command Code GOAT), with limited credits.

Through the API

$4.35 input, $8.70 output per 1M tokens.

Sources and data date

Show all 12 sources

Benchmark sources at a glance

Which independent evaluators tested this model. This is coverage, not a quality ranking.

No active result from the reviewed external evaluators can currently be attributed to this model. Other source evidence remains available below.

Without vendor claims and historical or disputed results; those stay labelled in the single values.

Type	Evidence and data date
Price	Base API prices without caching or batch discounts. Higher context tiers and other rates are listed under Costs. Source: Pricing · Xiaomi MiMo API pricing (retrieved October 25, 2026) · Xiaomi MiMo API pricing and models · Editorial description reviewed October 2, 2026
Research status	Research date October 15, 2026. 5 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication.
Unresolved	broader-independent-quality-benchmarks Note The exact UltraSpeed API route has separate Python micro-suite and ToneBench English long-form writing results. These specialized evaluations do not establish broader general-purpose capability or an immutable serving revision. A same-checkpoint statement alone does not transfer sibling-model results.
Additional source	Xiaomi MiMo model release (retrieved September 22, 2026; October 15, 2026) · Xiaomi MiMo V2.6 model release · Editorial description reviewed October 2, 2026
Additional source	Xiaomi MiMo V2.6 announcement (retrieved September 22, 2026; October 15, 2026) · Xiaomi MiMo V2.6 announcement · Editorial description reviewed October 2, 2026
Additional source	DataLLM Lab executed API benchmark (retrieved October 16, 2026; October 23, 2026)
Additional source	Towards AI ToneBench (retrieved October 17, 2026; October 28, 2026; October 30, 2026)
Additional source	Model metadata (retrieved October 16, 2026)
Additional source	OpenRouter: Popularity index (derived), CC BY 4.0
Additional source	Vercel AI Gateway: Popularity index (derived), CC BY 4.0
Additional source	Ollama: cumulative downloads for the popularity index
Additional source	Replicate: cumulative runs for the popularity index
Additional source	Google Trending Now: limited feed selection for Germany and the US
Additional source	Creative Commons Attribution 4.0