Xiaomi
MiMo-V2-Flash (Feb 2026)
Xiaomi MiMo-V2-Flash is a proprietary MoE model developed by Xiaomi, designed for extreme inference efficiency with 309B total parameters (15B active). By incorporating an innovative Hybrid attention architecture and multi-layer MTP inference acceleration, it ranks among the top 2 global open-source models across multiple Agent benchmarks.
Xiaomi model releasesModel Snapshot
TypeLanguage
ReleaseDec 17, 2025
Context262,144
Tagsreasoning, tool-use
Overall
Rank #67 across 526
41.5
Coding
Rank #107 across 436
33.5
Blended Price
Rank #46 across 357
$0.150/M
Strength Profile
Percentile score by analysis domain.
* Cost is inverted: lower input, output, and blended prices rank higher.
Benchmark Percentiles
Higher bars mean stronger relative placement.
Model Details
Runtime
Output Speed124.9 tok/s
First Token1.40s
Max Output32,000
Pricing
Blended$0.150/M
Input$0.100/M
All Benchmarks
| Metric | Domain | Value | Rank |
|---|---|---|---|
| Artificial Analysis Intelligence Index | overall | 41.5 | #67 |
| Artificial Analysis Coding Index | coding | 33.5 | #107 |
| GPQA | reasoning | 83.5% | #83 |
| Humanity's Last Exam | reasoning | 20.0% | #78 |
| SciCode |