OmniCompass-VL - Question Answering - Consistency Score: leaderboard
OmniCompass-VL - Question Answering - Consistency Score. Matched text-only/text-plus-image target specifications; consistency requires both target forms correct. 820 visual-language videos, 2460 task instances. Source-specific frame budgets retained in primary table; no fusion with single-condition accuracy. Metric: Consistency score (%). Primary published table; source configurations remain distinct.
Metric: Consistency score (%). Source: arxiv.org. Saturation forecast: Around 2029. 36 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.8 Flash | 69.1 | #21 |
| 2 | Kimi K3 | 67.6 | #30 |
| 3 | Qwen 3.8 Max | 67.2 | #31 |
| 4 | Gemini 3.1 Pro (Preview) | 65.7 | #56 |
| 5 | GPT-5.5 | 62.7 | #32 |
| 6 | GPT-5.6 Sol | 62.4 | #18 |
| 7 | Qwen 3.8 Flash | 62.4 | #45 |
| 8 | GPT-5.6 Terra | 62 | #42 |
| 9 | Qwen 3 VL 235B A22B Instruct | 61 | #267 |
| 10 | Qwen 3.5 397B A17B | 60.6 | #144 |
| 11 | GPT-6 Astra | 59.4 | #5 |
| 12 | GLM-5.3 Flash | 59.4 | #66 |
| 13 | GPT-5.6 Luna | 56.7 | #69 |
| 14 | DeepSeek V4.1 Flash | 56.6 | #36 |
| 15 | Qwen 3.8 27B | 55.7 | #96 |
No result here: #3 Claude Opus 5.5, #7 Gemini 4 Argon, #8 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=omnicompass-vl-question-answering-consistency-score · How It Works · Data refreshed daily, snapshot 2026-10-16.