SIBench: leaderboard
SIBench — visual-spatial intelligence suite aggregating spatial-reasoning tasks for VLMs; overall score.
Metric: Overall Score (%). Source: sibench.github.io. Saturation forecast: Around February 2027. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 63.41 | #98 |
| 2 | Gemini 2.5 Pro | 58.83 | #152 |
| 3 | InternVL3-78B | 51.97 | #352 |
| 4 | Gemini 2.5 Flash | 43.89 | #245 |
| 5 | GPT-4o Mini | 42.78 | #599 |
| 6 | Qwen 2.5 VL 7B | 41.72 | #643 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=sibench · How It Works · Data refreshed daily, snapshot 2026-10-20.