MiMo v2 Flash
Dec 2025
Older versionReplace withMiMo v2.6 FlashNewer version
Agnes 2.5 Pro BetaCheaper, as good
GPT-6 LunaBetter
Step 5 PreviewOpen weights
MiMo v2.6 FlashNewer versionAgnes 2.5 Pro BetaCheaper, as goodGPT-6 LunaBetterStep 5 PreviewOpen weightsMiMo v2 FlashCompare inon
Benchmarks of MiMo v2 Flash44
BenchmarkByScoreRank
Overall2
Knowledge8
Science4
Coding4
Agents1
Writing16
Documents2
Legacy · saturated or no longer reported7
No benchmark matches
Rankingin non-thinking mode?
Retired models are left out of the current rankings.
Alternatives?
ModelScoreCostPer taskWhat changes
CheaperAgnes 2.5 Pro Beta×2.2-14%$0.10 / $0.30BetterGPT-6 Lunalow+34%+14%$0.005Function calling only with none · Chat Completions+3
OpenDeepSeek V4.1 Flash 1017max×2.8+200%$0.26Memory (4-bit): 186 GB → 458 GB+5
CheaperLing 3.0 Flash VLthinkingOpen+53%-49%$0.06 / $0.18
CheaperSolar Mini 4+50%-50%$0.37ModelScoreCostPer taskWhat changes
Agnes 2.5 Pro Beta×2.2-14%$0.10 / $0.30Ling 3.0 Flash VLthinkingOpen+53%-49%$0.06 / $0.18Solar Mini 4+50%-50%$0.37Ling 3.0 Flash FinthinkingOpen+41%-49%$0.06 / $0.18Ling 3.0 FlashthinkingOpen+25%-49%$0.06 / $0.18ModelScoreCostPer taskWhat changes
GPT-6 Lunalow+34%+14%$0.005Function calling only with none · Chat Completions+3GPT-5.6 Lunalow+31%+157%$0.01MiMo v2.5thinkingOpen+36%same$0.019
Claude Haiku 5.5low+84%+14%$0.024Remove sampling parameters+5MiMo v2.5 ProthinkingOpen+62%+211%$0.054Memory (4-bit): 186 GB → 614 GBModelScoreCostPer taskWhat changes
DeepSeek V4.1 Flash 1017maxOpen×2.8+200%$0.26Memory (4-bit): 186 GB → 458 GB+5
GLM 5.3maxOpen×2.8+1,129%$2.01Memory (4-bit): 186 GB → 452 GB+4Step 5 PreviewthinkingOpen×2.7+714%$1.03Memory (4-bit): 186 GB → 363 GB+3GLM 5.3 FlashmaxOpen×2.6+36%$0.28Memory (4-bit): 186 GB → 193 GB+5Ling 3.1 FlashOpen×2.6+157%$1.00Memory (4-bit): 186 GB → 343 GBModelScoreCostPer taskWhat changes
MiMo v2.6 ProthinkingOpen×2.9—$0.13Memory (4-bit): 186 GB → 615 GB+4MiMo v2.6 FlashthinkingOpen×2.4—$0.062Pass back reasoning_content+3MiMo v2 Prothinking+78%—$0.43 / $0.87MiMo v2.5 ProthinkingOpen+62%—$0.054Memory (4-bit): 186 GB → 614 GBMiMo v2 Omni 0327+57%——MiMo v2 Omnithinking+49%—$0.14 / $0.28MiMo v2.5thinkingOpen+36%—$0.019MiMo v2.6 Pro UltraSpeedOpenn/a—$4.35 / $8.70Memory (4-bit): not verifiedAll Xiaomi models →Run it yourself
estimateFits on✗ 8 GB (phone, laptop)✗ 16 GB✗ 24 GB GPU✗ 48 GB GPU✗ 80 GB GPU
LicenceMITfree for commercial use
History
3 versionsBenchmarks
all categoriesChat & Knowledge
Arena Text · overall1,392
Arena Text · Chinese1,462.6
Arena Text · English1,429.6
Arena Text · French1,419.5
Arena Text · German1,384.2
Arena Text · Japanese1,325.5
Arena Text · Korean1,357.2
Arena Text · non-English1,391.6
Arena Text · Polish1,406.4
Arena Text · Russian1,387.4
Arena Text · Spanish1,420.7
AA Omniscience Index-48.5
GPQA Diamond65.6
Coding
WebDev Arena1,330
WebDev Arena · full cohort1,330.3
Agents & Tool Use
τ²-bench · AA83.9
Writing & Language
FAQ
How much does MiMo v2 Flash cost?›
$0.14 per 1M input tokens and $0.28 per 1M output tokens; cached input $0.003 per 1M.
Is MiMo v2 Flash free to use?›
There is no free API; the cheapest paid price is $0.14 per 1M input tokens. The weights are open, so you can run it yourself.
How can I use MiMo v2 Flash?›
Through the API of Xiaomi.
What is the context window of MiMo v2 Flash?›
256K tokens, with up to 66K output tokens.
When was MiMo v2 Flash released? Is there a newer version?›
Released 2025-12-16. It has been superseded by MiMo-V2-Flash (Feb 2026).
Is MiMo v2 Flash open source?›
Yes, the weights are open (MIT licence).
Can I run MiMo v2 Flash locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can MiMo v2 Flash read images?›
No, it accepts text only.
What is MiMo v2 Flash?›
MiMo v2 Flash is an open-weights language model from Xiaomi, released Dec 2025.
What are the best alternatives to MiMo v2 Flash?›
Agnes 2.5 Pro Beta (cheaper), GPT-6 Luna (better), DeepSeek V4.1 Flash 1017 (open weights).