OVO-S-Bench - Generative Spatial Reasoning: leaderboard
Level 3 of OVO-S-Bench (InternLM, 2026), a streaming spatial intelligence benchmark: multiple-choice accuracy in percent on generative spatial reasoning (simulation, consistency checks and route planning), with 128 frames sampled from the video before the query time. Higher is better.
Metric: Level 3 (generative spatial reasoning: simulation, consistency checks, route planning) multiple-choice accuracy (%) on the 1,680-question evaluation set, prefix-access mode (128 frames sampled uniformly from the video before the query timestamp, question given with the frames), mean of its three task families. Source: arxiv.org. Saturation forecast: Around 2029. 31 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 235B A22B | 61.2 | #479 |
| 2 | Qwen 3.5 397B A17B | 58.07 | #246 |
| 3 | Gemini 3.1 Pro (Preview) | 55.9 | #94 |
| 4 | Qwen 3 VL 4B Instruct | 54.54 | #1279 |
| 5 | Gemini 3.1 Flash Lite | 54.14 | #364 |
| 6 | Qwen 3.5 27B (Non-reasoning) | 52.39 | #359 (Qwen 3.5 27B) |
| 7 | Qwen 3 VL 32B Instruct | 51.94 | #585 |
| 8 | GPT-5.4 | 50.8 | #123 |
| 9 | Qwen 3.5 9B (Non-reasoning) | 50.2 | #803 (Qwen 3.5 9B) |
| 10 | Qwen 3.5 4B (Non-reasoning) | 49.25 | #1175 (Qwen 3.5 4B) |
| 11 | Grok 4.1 Fast (Non-reasoning) | 48.49 | #430 (Grok 4.1 Fast) |
| 12 | Qwen 2.5 VL 7B Instruct | 45.86 | #1833 |
| 13 | Gemma 4 26B A4B | 45.01 | #615 |
| 14 | Gemma 4 E4B | 42.8 | #2214 |
| 15 | Gemma 4 E2B | 39.34 | #3892 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=ovo-s-bench-generative-spatial-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-13.