DeepSeek V3.1 (Non-reasoning): benchmark results
DeepSeek V3.1 evaluated with reasoning disabled. Provider: DeepSeek. Released 2025-08-21. Access: Open.
Aggregate Rating 1599 ± 12, rank #719 of 2168 rated models, from 128 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| TravelEval - Cuisine Preference | 83.54 | Preference Achievement Sign for cuisine (PAS-C): plans honou | 100 |
| SalesLLM - Chinese (GPT-4o User) | 6.74 | SalesLLM benchmark score (1-10): 0.5 x buying-intent score ( | 96.4 |
| SalesLLM - Chinese (CustomerLM User) | 7.03 | SalesLLM benchmark score (1-10): 0.5 x buying-intent score ( | 92.9 |
| UGI Leaderboard | 51.4 | UGI Score | 91.5 |
| AgentNoiseBench (Search, Tool Noise) | 37 | Accuracy on 500 sampled 2WikiMultiHopQA and HotpotQA questio | 91.3 |
| Generalization V1 (Lechmazur) | 1.75 | Avg Rank (lower is better) | 88.1 |
| UGI - Writing | 51.06 | Writing Score | 87 |
| UGI - Natural Intelligence | 45.29 | NatInt Score | 86.7 |
| AA MMLU-Pro | 83.28 | Accuracy (%) | 83.9 |
| TravelEval - Accommodation Preference | 52.77 | Preference Achievement Sign for accommodation type (PAS-A): | 83.3 |
| TravelEval - Headcount Compliance | 64.64 | Headcount Compliance Sign (HCS): plans whose accommodation, | 83.3 |
| TravelEval - Intercity Time Deviation | 0.14 | Intercity Traffic Deviation (ITD): relative deviation of pla | 83.3 |
Interactive version: theaggregate.ai/model?slug=deepseek-v3-1-non-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-20.