VANTAGE-Bench - Semantic: leaderboard
VANTAGE-Bench vision-language video-understanding leaderboard — semantic accuracy.
Metric: Semantic (%). Source: huggingface.co. Saturation forecast: Around January 2027. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Astra | 85.08 | #6 |
| 2 | Gemini 3.6 Flash | 79.41 | #51 |
| 3 | GPT-5.6 Sol | 77.11 | #19 |
| 4 | Gemini 3.1 Pro (Preview) | 70.01 | #58 |
| 5 | Gemini 3.5 Flash Lite | 68.01 | #156 |
| 6 | Gemini 3.1 Flash Lite | 65.91 | #191 |
| 7 | Qwen 3 VL 32B Instruct | 65.67 | #283 |
| 8 | Qwen 3 VL 8B Instruct | 62.91 | #420 |
| 9 | Qwen 3.5 27B | 61.65 | #187 |
| 10 | Qwen 3.5 9B | 59.2 | #379 |
| 11 | Gemma 4 E2B (IT) | 36.92 | #781 |
No result here: #3 Claude Opus 5.5, #7 GPT-6.1 Sol, #8 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=vantage-bench-semantic · How It Works · Data refreshed daily, snapshot 2026-10-20.