DeepSeek V4 Pro (Non-reasoning) — benchmark results
DeepSeek V4 Pro non-reasoning evaluation mode. Provider: DeepSeek. Released 2026-04-23. Access: Open.
Unified ELO 1728 ± 27, rank #224 of 1776 rated models, from 36 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| UGI - Natural Intelligence | 59.21 | NatInt Score | 94.3 |
| UGI - Writing | 60.25 | Writing Score | 93.7 |
| UGI Leaderboard | 48.55 | UGI Score | 87.6 |
| AA TAU-2 Bench | 91.23 | Accuracy (%) | 86.1 |
| AA Omniscience - Software Engineering (SWE) - Java | 34 | Accuracy (%) | 83.5 |
| AA Terminal-Bench Hard | 36.36 | Accuracy (%) | 83.3 |
| AA Omniscience - Software Engineering (SWE) - PHP | 48 | Accuracy (%) | 82.7 |
| AA Omniscience - Business | 28.2 | Accuracy (%) | 82.4 |
| AA Omniscience - Software Engineering (SWE) - TypeScript | 47.78 | Accuracy (%) | 81.1 |
| AA Omniscience - Humanities & Social Sciences | 31.1 | Accuracy (%) | 80.6 |
| AA SciCode | 42.36 | Accuracy (%) | 80.4 |
| AA-Omniscience Accuracy | 30.77 | Accuracy (%) | 80.4 |
Interactive version: theaggregate.ai/model?slug=deepseek-v4-pro-non-reasoning · How the rankings work · Data refreshed daily, snapshot 2026-07-22.