DeepSeek-V3.2-Exp
deepseek-ai
- Arena score
- 1425
- Among open models
- #27of 144
- Among all on Arena
- #106
- Votes
- 8,991
Measured on DeepSeek-V3.2-Exp (Thinking), the served version, not a local build. LMArena, CC BY 4.0, published Oct 28, 2026.
Score by category
CategoryScoreArena rankVotes
- Medicine & health1468#42443
- Chinese1462#1081K
- Hard prompts, English1444#1022.1K
- Science1443#981.4K
- English1442#933.9K
- Spanish1441#82355
- German1440#75227
- Coding1439#1201.9K
- Software & IT1438#1243.2K
- Polish1435#73397
- French1434#112227
- Applied math1433#113403
- Hard prompts1429#1146.6K
- Expert1422#133425
- Math1422#122476
- Multi-turn1421#1162K
- Legal & government1420#131501
- Long prompts1417#1153.1K
- Excluding ties1414#1086.4K
- Business & finance1412#1212.2K
- Russian1412#113654
- Writing & literature1407#1012.7K
- Media & sports1407#792.2K
- Non-English1404#1255.1K
- Creative writing1403#921.7K
- Instructions1401#1293.4K
- Korean1370#112212
- Japanese1331#145148
BenchClub runs
Not runnable yet
Runs open once a DeepSeek-V3.2-Exp build is registered.