NEWHow AI models draw the world map · Gecko Tests
Beta
Home/Models/MiMo-V2-Flash
xiaomi logo

MiMo-V2-Flash

by xiaomi · Released Dec 2025

Open Source
81.7
avg score
Rank #17
Compare
Top 5% of all models
Context
262K tokens (~131 books)
Input $/1M
$0.09
Output $/1M
$0.29
Type
text
License
Open Source
Benchmarks
11 tested
Data as of
About

MiMo-V2-Flash is an open-source foundation language model developed by Xiaomi. It is a Mixture-of-Experts model with 309B total parameters and 15B active parameters, adopting hybrid attention architecture. MiMo-V2-Flash supports a...

Tested on 11 benchmarks · BenchGecko score 81.7. Top scores: Chatbot Arena Elo — Overall (1392.1%), Chatbot Arena Elo — Coding (1336.5%), OpenCompass — AIME2025 (92.9%).

Capabilities
coding
71.8
#14 globally
math
92.9
#7 globally
knowledge
61.9
#47 globally
speed
41.2
#58 globally
language
89.5
#17 globally
Benchmark Scores
Compare All
Tested on 11 benchmarks · Ranked across 6 categories
Score Distribution (all 312 models)
0255075100
▲ You are here
codingCompare coding →
OpenCompass — LiveCodeBenchV6

OpenCompass Live Code Bench v6. Fresh competitive programming problems to evaluate code generation without memorization.

71.8·
mathCompare math →
OpenCompass — AIME2025

OpenCompass evaluation on AIME 2025 problems. Tests mathematical reasoning on fresh competition problems.

92.9·
knowledgeCompare knowledge →
OpenCompass — MMLU-Pro

OpenCompass MMLU-Pro evaluation. Harder knowledge test with more answer choices.

83.1·
OpenCompass — GPQA-Diamond

OpenCompass evaluation of GPQA Diamond. PhD-level science questions from the hardest subset.

82.1·
OpenCompass — HLE

OpenCompass evaluation of Humanitys Last Exam. Expert-level cross-discipline knowledge test.

20.5·
Quick compare:vs PaLM 2-LOpenAI logovs GPT-5.6 SolOpenAI logovs GPT-5.4 Pro
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Model Family · xiaomi MiMo V2
Dec 2025
81.7
$0.09/M in262Kctx11 benchmarks
Mar 2026
$0.40/M in(+0.31)262Kctx3 benchmarks
Mar 2026
59.6-22.1
$1.00/M in(+0.60)1.0Mctx(+786K)13 benchmarks
Apr 2026
$0.14/M in(-0.86)1.1Mctx(+1K)3 benchmarks
Apr 2026
$0.43/M in(+0.29)1.1Mctx5 benchmarks
Sep 2026
$0.14/M in(-0.29)1.1Mctx8 benchmarks
Sep 2026
$0.43/M in(+0.29)1.1Mctx8 benchmarks
Sep 2026
$4.35/M in(+3.91)1.0Mctx(-1K)
See the full MiMo V2 family →
Similar Models
PaLM 2-L
Unknown
81.8n/a
OpenAI logo
GPT-5.6 Sol
OpenAI
80.9$2.00/1M
OpenAI logo
GPT-5.4 Pro
OpenAI
80.7$30.00/1M
Links
Info
xiaomiPricing explorerDevelopers · API
Research
Technical Report
Documentation
API DocsPlayground
Community
@xiaomi
Source Code
GitHubHugging Face
BenchGecko API
mimo-v2-flash
Specifications
  • Typetext
  • Context262K tokens (~131 books)
  • ReleasedDec 2025
  • LicenseOpen Source
  • StatusActive
  • Cost / Message~$0.000
Available On
xiaomi logoxiaomi$0.09
Categories
codingmathknowledgespeedlanguage
Learn More
context-windowtransformeropen weightstokens
Share & Export
Tweet
Related Models
PaLM 2-L
81.8
OpenAI logoGPT-5.6 Sol
80.9
OpenAI logoGPT-5.4 Pro
80.7
OpenAI logoGPT-5.2-Codex
80.7
MiMo-V2-Flash is an open-source text AI model by xiaomi, released in December 2025. It has an average benchmark score of 81.7. Context window: 262K tokens.

Related Models

PaLM 2-L · UnknownGPT-5.6 Sol · OpenAIGPT-5.4 Pro · OpenAIGPT-5.2-Codex · OpenAIClaude Opus 4.6 (Fast) · Anthropic

Benchmarks

Chatbot Arena Elo — OverallChatbot Arena Elo — CodingOpenCompass — AIME2025OpenCompass — IFEvalOpenCompass — MMLU-Pro

Related Pages

xiaomi · Providerxiaomi · EconomyAll ModelsCompare ModelsPricingDevelopers · APIContext Window · Glossary

Key facts · as of 2026-05-03

  • MiMo-V2-Flash by xiaomi. BenchGecko score 81.7, rank 17 of 312 scored models (normalized average of public benchmark scores).
  • List price $0.0900 input · $0.29 output per 1M tokens (as of 2026-05-03).

How to cite · data as of 2026-05-03

MiMo-V2-Flash · benchmarks, pricing and providers. BenchGecko, data as of 2026-05-03. https://benchgecko.ai/model/mimo-v2-flash

Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP