DeepSeek V3.2 (Non-reasoning): benchmark results
DeepSeek V3.2 evaluated with reasoning disabled. Provider: DeepSeek. Released 2025-12-01. Access: Open.
Unified ELO 1575 ± 1, rank #534 of 1761 rated models, from 61 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CritPt | 90 | Accuracy (self-reported) | 97.7 |
| UGI Leaderboard | 55.75 | UGI Score | 96.4 |
| UGI - Writing | 56.3 | Writing Score | 91 |
| UGI - Natural Intelligence | 47.85 | NatInt Score | 89.5 |
| AA Terminal-Bench Hard | 32.58 | Accuracy (%) | 76.6 |
| UGI - Willingness (W/10) | 7.2 | W/10 Score | 73.9 |
| AA TAU-2 Bench | 78.95 | Accuracy (%) | 69.1 |
| AA Omniscience - Humanities & Social Sciences | 25.9 | Accuracy (%) | 66.3 |
| AA Omniscience - Law | 17.3 | Accuracy (%) | 65.1 |
| AA Omniscience - Software Engineering (SWE) | 33.7 | Accuracy (%) | 64.5 |
| AA Global-MMLU-Lite - English | 91.08 | Accuracy (%) | 64 |
| Artificial Analysis Intelligence Index | 18.27 | Intelligence Index | 63.1 |
Interactive version: theaggregate.ai/model?slug=deepseek-v3-2-non-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-05.