DeepSeek-V3.2-Exp

DeepSeek
Arena score
1425
Among open models
#28of 148
Among all on Arena
#105
Votes
8,963

Measured on DeepSeek-V3.2-Exp (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#9Medicine & health

CategoryScore
  • Medicine & health1469
  • Chinese1462
  • Science1444
  • Hard prompts, English1443
  • English1441
  • Spanish1440
  • German1440
  • Coding1439
  • Software & IT1438
  • Polish1435
  • French1433
  • Applied math1433
  • Hard prompts1429
  • Expert1423
  • Multi-turn1421
  • Legal & government1421
  • Math1420
  • Long prompts1418
  • Excluding ties1414
  • Business & finance1412
  • Russian1412
  • Media & sports1407
  • Writing & literature1407
  • Non-English1404
  • Creative writing1403
  • Instructions1401
  • Korean1371
  • Japanese1332

Score by size

Plot against
Best for its size

About as strong, smaller

Compare

Built on it

Quantizations
3
Fine-tunes
13
Adapters
3
Merges
0

Ways to run it

Most downloaded

BenchClub runs

Not runnable yet

Runs open once a DeepSeek-V3.2-Exp build is registered.

More from DeepSeek

See more

Other >300B models

See more