DeepSeek V4 Pro (Reasoning): benchmark results
Provider: DeepSeek. Released 2026-04-23. Access: Open.
Unified ELO 1680 ± 1, rank #156 of 1761 rated models, from 9 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| UGI Leaderboard | 62.26 | UGI Score | 99.1 |
| UGI - Writing | 68.4 | Writing Score | 97.7 |
| UGI - Natural Intelligence | 67.03 | NatInt Score | 96.6 |
| Wolfram LLM Benchmarking Project | 62.7 | Correct Functionality (%) | 91.4 |
| AtmosCoder-Bench | 93.3 | Accuracy (%, mean of 3 runs) | 80 |
| GRIPS | 92.9 | Accuracy (%) | 76.9 |
| GRIPS-hard | 67.3 | Accuracy (%) | 76.9 |
| SpeechMap Compliance | 64.8 | % Requests Completed | 54.7 |
| UGI - Willingness (W/10) | 3.2 | W/10 Score | 24.6 |
Interactive version: theaggregate.ai/model?slug=deepseek-v4-pro-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-05.