DeepSeek-V4-Flash-Vision-Exp: benchmark results
DeepSeek's open-weight DeepSeek-V4-Flash-Vision-Exp (August 2026), the DeepSeek-V4 family's first experimental multimodal model, built on the V4 Flash architecture with added vision capabilities. Provider: DeepSeek. Released 2026-08-21. Access: Open.
Aggregate Rating 1782 ± 16, rank #79 of 1635 rated models, from 146 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Nejumi 4 - jaster (2-shot) - JCoLA (out-of-domain) | 89 | Exact match (%) | 94.7 |
| Nejumi 4 - jaster (0-shot) - CommonsenseMoralJA | 97 | Exact match (%) | 93.6 |
| Nejumi 4 - ARC-AGI-1 | 98 | Accuracy (%) | 93.3 |
| Korean CSAT 2026 (Easy Mode) - Korean | 100 | Points (out of 100) | 92.6 |
| Korean CSAT 2026 (Easy Mode) - English | 100 | Points (out of 100) | 90.5 |
| Nejumi 4 - BFCL - Relevance Detection | 88.89 | Accuracy (%) | 89.4 |
| LLM Stats Score | 44.71 | LLM Stats Score (conservative rating) | 89.2 |
| Nejumi 4 - HLE (Japanese) - Calibration Error | 53 | Calibration error (%) | 87.9 |
| Nejumi 4 - jaster (0-shot) - NIILC | 57.72 | Character F1 (x100) | 87.9 |
| Nejumi 4 - HLE (Japanese) - Accuracy | 38.14 | Accuracy (%) | 87.2 |
| Nejumi 4 - GLP - Specialized Knowledge | 72.55 | Score (%) | 86.5 |
| Nejumi 4 - jaster (0-shot) - KUCI | 87 | Exact match (%) | 86.5 |
Interactive version: theaggregate.ai/model?slug=deepseek-v4-flash-vision-exp · How It Works · Data refreshed daily, snapshot 2026-10-15.