DeepSeek V3.1 (Non-reasoning): benchmark results
DeepSeek V3.1 evaluated with reasoning disabled. Provider: DeepSeek. Released 2025-08-21. Access: Open.
Unified ELO 1547 ± 1, rank #646 of 1761 rated models, from 23 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| UGI Leaderboard | 51.4 | UGI Score | 92.2 |
| UGI - Writing | 51.06 | Writing Score | 88.7 |
| UGI - Natural Intelligence | 45.29 | NatInt Score | 88.3 |
| AA Terminal-Bench Hard | 24.24 | Accuracy (%) | 64.5 |
| AA Omniscience - Humanities & Social Sciences | 25.45 | Accuracy (%) | 64.1 |
| AA Omniscience - Law | 16.42 | Accuracy (%) | 63.5 |
| AA Omniscience - Business | 19.82 | Accuracy (%) | 61 |
| AA Omniscience - Software Engineering (SWE) | 32.2 | Accuracy (%) | 61 |
| Artificial Analysis Intelligence Index | 14.82 | Intelligence Index | 57.8 |
| AA-Omniscience Accuracy | 23.13 | Accuracy (%) | 57.3 |
| AA GPQA Diamond | 73.54 | Accuracy (%) | 54.2 |
| AA Omniscience - Health | 19.7 | Accuracy (%) | 46.6 |
Interactive version: theaggregate.ai/model?slug=deepseek-v3-1-non-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-05.