DeepSeek V4.1 Flash (1017) (Low): benchmark results
Provider: DeepSeek. Released 2026-10-17. Access: Open.
Aggregate Rating 1854 ± 27, rank #76 of 2170 rated models, from 15 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LisanBench | 0.31 | Mean Path Length / Current Maximum | 90.5 |
| Bug Hunt Bench - LMS | 19.8 | Planted Bugs Fixed (out of 60) | 84.3 |
| Bug Hunt Bench - VS Code Extension | 16.4 | Planted Bugs Fixed (out of 45) | 84.3 |
| ARC-AGI-2 | 73.61 | Accuracy (%) | 82.1 |
| Bug Hunt Bench | 34.8 | Planted Bugs Fixed (out of 105) | 81.8 |
| ARC-AGI-1 | 91.5 | Accuracy (%) | 77.6 |
| Fangcun DeepSeek 1017 Report - NCRI reasoning-effort comparison | 98.5 | Accuracy (%) | 72.2 |
| CursorBench 4.0 | 44.9 | Score (%) | 65.1 |
| EnigmaForge | 44.2 | Task success (%) over 600 procedurally generated story insta | 59.3 |
| o11y-bench - Pass^3 | 79.69 | Tasks passed on all three attempts, Pass^3 (%) | 59.3 |
| Argo-Bench | 24.75 | Mean task score (0-100) over 210 enterprise data-science tas | 51.7 |
| o11y-bench - Pass@3 | 93.75 | Tasks passed on at least one of three attempts, Pass@3 (%) | 41.4 |
Interactive version: theaggregate.ai/model?slug=deepseek-v4-1-flash-1017-low · How It Works · Data refreshed daily, snapshot 2026-11-05.