DeepSeek-V3.1-Terminus
deepseek-ai
- Arena score
- 1419
- Among open models
- #34of 144
- Among all on Arena
- #123
- Votes
- 3,379
Measured on DeepSeek-V3.1-Terminus (Thinking), the served version, not a local build. LMArena, CC BY 4.0, published Oct 28, 2026.
Score by category
CategoryScoreArena rankVotes
- Medicine & health1468#43227
- Legal & government1458#57258
- Science1451#78571
- Russian1436#73283
- Hard prompts, English1432#124729
- English1428#1231.5K
- Polish1427#90205
- Hard prompts1426#1231.7K
- Software & IT1426#1461.1K
- Coding1425#145633
- Long prompts1421#111680
- Business & finance1415#114692
- Multi-turn1411#137571
- Writing & literature1409#96761
- Non-English1407#1132K
- Excluding ties1406#1232.4K
- Instructions1405#120858
- Creative writing1403#90446
- Math1402#161216
- Media & sports1392#115609
BenchClub runs
Not runnable yet
Runs open once a DeepSeek-V3.1-Terminus build is registered.