All models › Compare

kimi-k2.5-thinking vs MiMo-V2.6-Flash

MiMo-V2.6-Flash is ahead on Arena (1455 vs 1445): in a blind vote people prefer it about 51% of the time.

kimi-k2.5-thinking1445Arena chat · #62 of 416
MiMo-V2.6-Flash1455Arena chat · #45 of 416

Arena scores by category

Categorykimi-k2.5-thinkingMiMo-V2.6-FlashAhead
Chat14451455MiMo-V2.6-Flash
Vision12691259kimi-k2.5-thinking

Scores from blind human votes on Arena. A gap under 10 points is too small to call.

Size and where to use them free

kimi-k2.5-thinkingMiMo-V2.6-Flash
Made bymoonshotxiaomi
Sizenot public310.8 B (17.4 B active)
4-bit download—190 GB
Use it freeModified MIT licensedownload and run at home · MIT license

Speed at home

Hardwarekimi-k2.5-thinkingMiMo-V2.6-Flash
RTX 3090online onlydoes not fit
RTX 4090online onlydoes not fit
RTX 5090online onlydoes not fit
RX 7900 XTXonline onlydoes not fit
Mac M4 Max 128 GBonline onlydoes not fit
Ryzen AI Max+ 395 128 GBonline onlydoes not fit

Estimates at 4-bit with an 8K context, 64 GB of system RAM for offload, calibrated on 44 real llama.cpp tests.

Try them yourself

Ask both the same question side by side in Try & compare, or vote blind between free models in Free Arena.

More comparisons

Share this page:XLinkedInRedditWhatsApp