TOC-Bench - Conditional State: leaderboard
The conditional state questions of TOC-Bench (2026), video questions grounded in per-frame object tracks and kept only when not answerable from text alone, single frames or shuffled frames: whether an object is fully visible, partly hidden, fully hidden or absent at the moment of another event. Accuracy in percent; higher is better.
Metric: Conditional State accuracy (%) on TOC-Bench (2,323 human-verified questions over 1,951 videos from Charades, Perception Test, OVIS and MOSE, grounded in per-frame object tracks and kept only when not answerable from text alone, single frames or shuffled frames), mostly 32 uniformly sampled frames, deterministic scoring of multiple-choice, statement-pair, numeric and ordering formats; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 21 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Kimi K2.6 | 56.5 | #192 |
| 2 | Grok 4.3 | 51.8 | #342 |
| 3 | Gemini 3.1 Pro (Preview) | 50 | #99 |
| 4 | Qwen 2.5 VL 7B | 43.9 | #1862 |
| 5 | Seed 2.0 Lite | 43.5 | #300 |
| 6 | MiMo-V2-Omni | 42.1 | #465 |
| 7 | GPT-5.5 | 41.7 | #54 |
| 8 | Qwen 3 VL 8B (Thinking) | 40.6 | #910 |
| 9 | Qwen 2.5 VL 72B | 40.6 | #781 |
| 10 | VideoChat-R1.5-7B | 40.6 | #1254 |
| 11 | Gemini 3.1 Flash Lite (Preview) | 39.6 | #417 |
| 12 | GLM-5V Turbo | 36.7 | #360 |
| 13 | GPT-5.4 Mini | 36.3 | #406 |
| 14 | Video-R1 | 35.3 | #1379 |
| 15 | InternVL3-8B | 31.7 | #1782 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=toc-bench-conditional-state · How It Works · Data refreshed daily, snapshot 2026-11-21.