Kimi-K2.5

Moonshot AI
Arena score
1445
Among open models
#13of 148
Among all on Arena
#67
Votes
73,820

Measured on Kimi K2.5 (Thinking), the served version, not a local build. LMArena, CC BY 4.0, published Nov 11, 2026.

Score by category

Top 10 among open models in#9Coding#10Applied math

CategoryScore
  • Chinese1493
  • Coding1484
  • Applied math1473
  • Software & IT1471
  • Math1469
  • Expert1465
  • Hard prompts, English1460
  • Science1458
  • French1454
  • English1453
  • Hard prompts1453
  • Medicine & health1449
  • Spanish1449
  • Polish1445
  • Multi-turn1444
  • Long prompts1444
  • Excluding ties1442
  • German1441
  • Legal & government1441
  • Russian1435
  • Non-English1432
  • Instructions1431
  • Business & finance1427
  • Writing & literature1425
  • Japanese1422
  • Creative writing1422
  • Media & sports1417
  • Korean1410

Score by size

Plot against
Best for its size

About as strong, smaller

Compare

Discussion

Built on it

Quantizations
44
Fine-tunes
47
Adapters
29
Merges
1

Ways to run it

Most downloaded

BenchClub runs

Not runnable yet

Runs open once a Kimi-K2.5 build is registered.

More from Moonshot AI

See more

Other >300B models

See more