DeepSeek V4.1 Flash (High): benchmark results
Provider: DeepSeek. Released 2026-09-10. Access: Open.
Aggregate Rating 1791 ± 12, rank #157 of 2170 rated models, from 71 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Vals AI SkillsBench | 69.8 | Accuracy (%) | 100 |
| Vals Inference Provider - First Party - Vibe Code Bench | 85.49 | Accuracy (%) | 100 |
| ParseBench - Semantic Formatting | 73.48 | Semantic Formatting Score | 88.3 |
| Vals AI Vibe Code Bench | 84.74 | Accuracy (%) | 85.4 |
| Vals AI Code Migration | 45.62 | Accuracy (%) | 79.7 |
| Chess Bench LLM | 1277 | Lichess-Classical-calibrated Glicko-2 rating (native signed | 79.5 |
| Vals AI CyberBench | 73.69 | Accuracy (%) | 79.5 |
| ARC-AGI-2 | 67.5 | Accuracy (%) | 77.2 |
| Vals AI Terminal-Bench 2.1 | 74.53 | Accuracy (%) | 76.7 |
| SweetBench - Glucose variability subscore | 80.64 | Control score (0–100) | 75 |
| Vals AI MedScribe | 85.5 | Accuracy (%) | 74.5 |
| ObviousBench | 97.22 | Answer pass³ (%) | 71.4 |
Interactive version: theaggregate.ai/model?slug=deepseek-v4-1-flash-high · How It Works · Data refreshed daily, snapshot 2026-11-05.