xiaomi · Open weights
MiMo V2.6 Pro UltraSpeed
MiMo V2.6 Pro UltraSpeed is the latency-optimized serving mode for Xiaomi's flagship MiMo V2.6 Pro checkpoint. It preserves Pro-level quality while delivering up to 20x faster output for real-time coding, interactive agents, and other latency-sensitive workflows, with native text, image, video, and audio understanding and a 1M-token context window.
- Reasoning
- Tools
- Vision
- Structured output
- Audio input
- Video input
- Input
- Text, Images, Video, Audio
- Output
- Text
- Reasoning effort
- none · high
- Discovery tier
- Open weights
Pricing
USD per 1M tokens- Input
- $4.35
- Output
- $8.70
- Cache read
- $0.036
- Cache write
- $0.00
Reference prices as published by the model provider. API example
Specifications
- Context window
- 1M
- Max output
- 131.1K
- Average output (7 days)
- 1K tokens
- Longer than 70% of 468 models
Side by side
MiMo V2.6 Pro UltraSpeed compared
Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.
| MiMo V2.6 Pro UltraSpeedxiaomiThis model | MiMo V2.6 Flash AbliteratedxiaomiRemove | |
|---|---|---|
| Provider | xiaomi | xiaomi |
| Discovery tier | Open weights | Open weights |
| Added | Sep 21, 2026 | Sep 25, 2026 |
| Limits | ||
| Context window | 1.05M | 1.05M |
| Max output | 131.07KBest | 65.54K |
| Average output | 1K tokens | 1.3K tokens |
| Pricing, USD per 1M tokens | ||
| Input | $4.35 | $0.50Best |
| Output | $8.70 | $1.50Best |
| Cache read | $0.036Best | $0.28 |
| Cache write | $0.00 | — |
| Performance | ||
| Throughput | 263.0 tok/sBest | 164.5 tok/s |
| Time to first token | 11.1 s | 1.5 sBest |
| Request success (24h) | 100.0%Best | 98.4% |
| Benchmark indexes | ||
| Intelligence Index | 46.3 | — |
| Capabilities | ||
| Reasoning | Yes | Yes |
| Tools | Yes | Yes |
| Vision | Yes | Yes |
| Structured output | Yes | No |
| Audio input | Yes | No |
| Video input | Yes | No |
| Reasoning effort | none, high | none, high |
Public evaluations
Benchmarks
Independent evaluation results, with how each score ranks among the models in the Noumen catalog.
- MiMo V2.6 Pro UltraSpeed
- MiMo V2.6 Flash Abliterated
Intelligence Index
- MiMo V2.6 Pro UltraSpeed46.3
- MiMo V2.6 Flash Abliterated—
MiMo V2.6 Pro UltraSpeed: Better than 92% of 326 scored models
Agentic work
AutomationBench-AA
Workflow automation, penalized for guardrail violations
- MiMo V2.6 Pro UltraSpeed: 58.6%
- MiMo V2.6 Flash Abliterated: —
AutomationBench-AA tasks completed
Workflows finished end to end without a guardrail violation
- MiMo V2.6 Pro UltraSpeed: 25.4%
- MiMo V2.6 Flash Abliterated: —
AA-Briefcasev1.1
Agentic knowledge work, rated head to head
- MiMo V2.6 Pro UltraSpeed: 1,512 Elo
- MiMo V2.6 Flash Abliterated: —
GDPval-AAv2.1
Economically valuable occupational tasks, rated head to head
- MiMo V2.6 Pro UltraSpeed: 1,683 Elo
- MiMo V2.6 Flash Abliterated: —
Document reasoning
GDP.pdf
Professional reasoning over PDF documents, all-pass rate
- MiMo V2.6 Pro UltraSpeed: 19.2%
- MiMo V2.6 Flash Abliterated: —
AA-LCRv1.1
Reasoning across long documents
- MiMo V2.6 Pro UltraSpeed: 86.3%
- MiMo V2.6 Flash Abliterated: —
MLCR-AA
Long-context reasoning over medical records
- MiMo V2.6 Pro UltraSpeed: 12.8%
- MiMo V2.6 Flash Abliterated: —
Reasoning and knowledge
Humanity’s Last Exam
Expert-level questions across academic fields
- MiMo V2.6 Pro UltraSpeed: 49.4%
- MiMo V2.6 Flash Abliterated: —
Coding
Terminal-Benchv4.0
Practical software engineering in a terminal
- MiMo V2.6 Pro UltraSpeed: 34.8%
- MiMo V2.6 Flash Abliterated: —
SciCode
Scientific computing in Python
- MiMo V2.6 Pro UltraSpeed: 60.9%
- MiMo V2.6 Flash Abliterated: —
Legacy evaluations
AA-LCR (unversioned)
Long-context reasoning, earlier grading
- MiMo V2.6 Pro UltraSpeed: 86.3%
- MiMo V2.6 Flash Abliterated: —
Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis
Live routing data
Performance and reliability
Recent averages across requests routed for this model. Your own speed depends on prompt length, reasoning effort and load.
- Throughput
- 263.0 tok/s
- Faster than 91% of 254 models
- Time to first token
- 11.1 s
- Quicker than 7% of 254 models
- Prompt cache hits
- 67.0%
- About $1.46 per 1M input tokens with caching
- Request success (24h)
- 100.0%
- Reliable
Cache hits are the share of input tokens recently served from a prompt cache across all requests. They are not a forecast of your own hit rate.
- Healthy (98% or more)
- Degraded (90–98%)
- Down (under 90%)
- No requests
Developers
Use it through the API
Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.
curl https://www.noumen.xyz/api/v1/chat/completions \
-H "Authorization: Bearer $NOUMEN_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "xiaomi/mimo-v2.6-pro-ultraspeed",
"reasoning_effort": "none",
"messages": [{ "role": "user", "content": "Explain this stack trace." }]
}'import os
from openai import OpenAI
client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
model="xiaomi/mimo-v2.6-pro-ultraspeed",
messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)xiaomi
Related models
Other xiaomi models, starting with the same family.