DeepSeek V4.1 Flash (Low): benchmark results
Provider: DeepSeek. Released 2026-09-10. Access: Open.
Aggregate Rating 1793 ± 33, rank #155 of 2169 rated models, from 16 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ARC-AGI-1 | 90.5 | Accuracy (%) | 74 |
| ARC-AGI-2 | 60.56 | Accuracy (%) | 70.7 |
| EnigmaForge | 48.7 | Task success (%) over 600 procedurally generated story insta | 62.5 |
| NonoBench | 73.33 | Standard accuracy (%), correct / non-failed attempts; versio | 57.8 |
| QuantArena - Iterative - Research Period | 48.46 | Score (%) | 45.9 |
| EnigmaForge - Intuition | 18 | Task success (%) on implicit-condition instances, where the | 45.8 |
| ARC-AGI-3 | 0.35 | Accuracy (%) | 35.8 |
| Argo-Bench | 20.42 | Mean task score (0-100) over 210 enterprise data-science tas | 35.7 |
| Argo-Bench - Solved | 8.57 | Share of the 210 tasks solved (%), a task counting as solved | 28.6 |
| QuantArena - One-shot - Research Period | 34.21 | Score (%) | 27 |
| EnigmaForge - Fact F1 | 78.7 | World-fact recovery F1 (%) over 600 instances, all items | 25 |
| QuantArena - Iterative - Overall | 36.45 | Score (%) | 24.3 |
Interactive version: theaggregate.ai/model?slug=deepseek-v4-1-flash-low · How It Works · Data refreshed daily, snapshot 2026-10-23.