DeepSeek V4 Flash (High): benchmark results
Provider: DeepSeek. Released 2026-04-24. Access: Open.
Aggregate Rating 1692 ± 13, rank #382 of 2170 rated models, from 87 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AutoRO-Bench (AutoREM) | 97.4 | Accuracy (%; share of the 64 in-distribution Random robust-o | 100 |
| tau-Rec - Pass^4 | 38.3 | pass^4 (0-1 scaled to %), the probability that the agent suc | 100 |
| SQBench - L3 Business Scenarios | 23.3 | Strict Pass@1 (%; the 60 L3 domain business-scenario tasks o | 88.5 |
| tau-Rec | 56 | pass^1 (0-1 scaled to %), the mean strict success rate per t | 87.5 |
| tau-Rec - Constraint Score | 58.6 | Mean constraint sub-score (0-1 scaled to %), the partial-cre | 87.5 |
| InfiniteBM Heads-Up No-Limit Hold'em | 1359.06 | Arena Elo (elo) | 84.8 |
| WebDev Arena (Content Creation Tools) | 1561 | Arena Score | 80.1 |
| WebDev Arena (Brand & Marketing) | 1607 | Arena Score | 79.9 |
| WebDev Arena (Consumer Product) | 1562 | Arena Score | 79.9 |
| WebDev Arena (Data & Analytics) | 1537 | Arena Score | 77.3 |
| Chatbot Arena (Code) | 1582 | Elo | 77 |
| WebDev Arena (Frontend) | 1585 | Arena Score | 76.4 |
Interactive version: theaggregate.ai/model?slug=deepseek-v4-flash-high · How It Works · Data refreshed daily, snapshot 2026-11-05.