CaST-Bench - QA Accuracy: leaderboard
Evaluates causal chain-grounded spatio-temporal reasoning for video question answering on CaST-Bench: 2,066 questions over 1,015 videos whose causal chains are annotated as temporal segments with per-second bounding boxes. Predicted evidence instances are greedily matched one-to-one against ground-truth instances by coverage-aware spatio-temporal IoU (vIoU*), the temporal IoU multiplied by the mean per-second spatial IoU over the overlapping seconds. QA Accuracy is the share of questions whose predicted multiple-choice option matches the ground-truth option, independent of evidence localisation. This is the open-submission Space leaderboard for CaST-Bench v1; published fractions are converted to percentages. This series reports CaST-Bench - QA Accuracy using Accuracy (%).
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Around February 2027. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 76.25 | #64 |
| 2 | Claude Sonnet 4.6 | 75 | #94 |
| 3 | Gemini 3.5 Flash | 72.5 | #66 |
| 4 | Claude Opus 4.7 | 70 | #51 |
| 5 | Qwen 3 VL 8B Instruct | 61.25 | #419 |
| 6 | Gemma 4 26B A4B (IT) | 61.25 | #293 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=cast-bench-qa-accuracy · How It Works · Data refreshed daily, snapshot 2026-11-05.