DeepSeek V4 Pro (0423): benchmark results
Provider: DeepSeek. Released 2026-04-24.
Aggregate Rating 1685 ± 38, rank #418 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenRouter Tau2-Bench Airline | 77.2 | Accuracy (%) | 88.8 |
| OpenRouter GPQA Diamond | 87.1 | Accuracy (%) | 67.4 |
| ObviousBench | 79.86 | Answer pass³ (%) | 30.3 |
| Convex Coding Evals — No Guidelines (111 tasks) | 34.23 | Pass rate (%) | 23.3 |
| KiloBench - Terminal Bench 2.0 | 44.04 | Task completion in Kilo's agent harness, mean of 5 attempts | 16.7 |
| SnakeBench | 8.6 | TrueSkill Rating | 3.4 |
Interactive version: theaggregate.ai/model?slug=deepseek-v4-pro-0423 · How It Works · Data refreshed daily, snapshot 2026-11-08.