Last verified
10X V2.6-PRO PRICEUP TO 20X OUTPUT SPEEDSAME MODEL AS V2.6-PRONO BATCH API1M CONTEXT

MiMo-V2.6-Pro-UltraSpeed API Pricing

The same model as MiMo-V2.6-Pro, served for latency: Xiaomi promises up to 20x the output speed with no loss of capability, and charges exactly 10x on every line. That is $4.35/M input on a cache miss, $0.036/M on a hit and $8.70/M output. The invite-only V2.5-Pro-UltraSpeed it replaces cost 3x its base model, so speed got more than three times dearer relative to the model underneath. There is no Batch API, and rate limits are set by arrangement with Xiaomi.

Input - per 1M tokens
$4.35/M
V2.6-Pro $0.435/M cache miss
Output - per 1M tokens
$8.70/M
V2.6-Pro $0.87/M
Cached input - per 1M tokens
$0.036/M
Cache writes free for now 99% off
Effective - agentic blend
$1.44/M
92/8 split - 82% cache
§ 01 / TERMINAL

Run the numbers.

Calculator pre-loaded with Xiaomi's overseas USD list for MiMo-V2.6-Pro-UltraSpeed. Every result is exactly 10x what the same work costs on MiMo-V2.6-Pro; the difference buys wall-clock time, not a better answer.

$ /mo
Workload split
Prompt cache hit rate
Tokens you can process
—
Words equivalent (English)
—
Effective rate
—
Open full calculator (all models · share URL · CSV) →
§ 02 / SCENARIOS

Real-world presets.

LIVE CODING

Inline code completion

$0.017/edit
12,000 in - 800 out~5,984 units/$100
TRADING

News-to-signal analysis

$0.010/event
6,000 in - 600 out~9,905 units/$100
RISK

Transaction fraud check

$0.00673/check
4,000 in - 400 out~14,858 units/$100
VOICE AGENT

Real-time support turn

$0.00667/turn
5,000 in - 300 out~14,986 units/$100
§ 03 / TOKENIZER

Paste text. See tokens. See cost.

What's a token →
Estimate · xiaomi-tokenizer-estimate · ≈3.85 chars/token Auto-counts as you type

This is a chars-per-token approximation, not a real tokenizer. Actual tokens vary by language, code density, and tool-call overhead — counts are typically ±10–20% off for English prose, more for code or non-Latin scripts. For exact billing, use the vendor's official tokenizer.

Characters 402
Words 64
Tokens (estimated) 104 tokens
Cost as input · uncached $0.00045 USD
Cost as output · uncached $0.0009 USD
Cost as cached input $0.000004 USD
§ 04 / SHELF

Up against the shelf.

All models →
Model Input /M Output /M Effective blended Context Best for
MiMo-V2.6-Pro-UltraSpeed Current $4.35 cache $0.036 $8.70 $1.44 agentic 92/8 1M Latency-bound work where output speed is the product
MiMo-V2.6-Pro $0.435 cache $0.0036 $0.87 $0.144 −10× 1M The same model at a tenth of the price
MiMo-V2.6-Flash $0.14 cache $0.0028 $0.28 $0.0477 −30× 1M Cheap and fast enough for most high-volume calls
MiMo-V2.5-Pro-UltraSpeed $1.30 cache $0.0108 $2.61 $0.433 withdrawn 1M What this model replaces, for the record
GPT-6.1 Sol $2.00 cache $0.10 $10.00 $1.29 −11% 1.05M Closed frontier model with a paid fast mode
Claude Sonnet 5.5 $2.00 cache $0.10 $10.00 $1.29 −11% 1M Closed frontier agent work

Frequently asked.

What the speed costs, what you get for it, and when the ordinary V2.6-Pro is the better buy.

Q · 01 How much does MiMo-V2.6-Pro-UltraSpeed cost? +
$4.35/M input on a cache miss, $0.036/M on a cache hit and $8.70/M output on Xiaomi's overseas list, or $1.44/M under our 92/8 agentic blend with 82% cache hits. There is no Batch API for this model, and cache writes are listed as limited-time free.
Q · 02 Is it a different model from MiMo-V2.6-Pro? +
No. Xiaomi describes it as the UltraSpeed mode of V2.6-Pro with the flagship's full capability and "no intelligence traded for speed", served for up to 20x the output speed. It has its own model id, mimo-v2.6-pro-ultraspeed, and its own price, which is why it has its own page. Same inputs (text, images, video, audio), same 1M context, same 128K output limit.
Q · 03 How does the price compare with MiMo-V2.6-Pro? +
Exactly 10x on every line at launch: $4.35 against V2.6-Pro's $0.435 input, $8.70 against $0.87 output, and the same ratio on cache hits. V2.6-Pro also has a Batch API at half its rate, so for work nobody watches in real time the gap is 20x.
Q · 04 Why did the speed premium go from 3x to 10x? +
The invite-only MiMo-V2.5-Pro-UltraSpeed listed at $1.305 / $0.0108 / $2.61, exactly 3x V2.5-Pro, and was pitched at a 1,000 tokens/s peak. The V2.6 version is pitched as up to 20x V2.6-Pro's output speed and priced at 10x. Xiaomi has not published tokens-per-second figures for the new model, so there is no way yet to say whether the price per unit of speed went up or down.
Q · 05 When is it worth paying for? +
When waiting for the answer costs more than the tokens: Xiaomi's own examples are trading signals from breaking news, fraud checks that must finish before settlement, and code completion that keeps pace with typing. For batch jobs, reports and background agents it buys nothing that V2.6-Pro does not deliver at a tenth of the price.
Q · 06 What are the rate limits? +
Xiaomi publishes none. Where the V2.6-Pro and Flash pages list 100 requests and 10M tokens per minute, this one says to contact Xiaomi for customised service. Plan capacity with the vendor before building a product on it.
Y
Reviewed by Yaroslav Vikhariev Founder · AI//COST · Prices checked by hand against mimo.mi.com · Last verified Sep 24, 2026
Methodology Report a correction More by Y.V.