Cambrian-S Release - Percept. Test †: leaderboard
Cambrian-S release paper; final Stage4 checkpoint. Benchmark-specific sampling of Section3 retained, dagger-marked video scores use paper protocol; no pretrained, intermediate-stage or LFP checkpoint merge. Metric: Accuracy (%). Higher is better. Source: https://arxiv.org/html/2511.04670v1
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Cambrian-S-7B | 69.9 | |
| 2 | LLaVA-Video-7B | 67.9 | |
| 3 | Qwen2.5-VL-3B | 66.9 | |
| 4 | Cambrian-S-3B | 65.9 | |
| 5 | Cambrian-S-1.5B | 63.2 | |
| 6 | LLaVA-OneVision-7B | 57.1 | |
| 7 | Cambrian-S-0.5B | 56 | |
| 8 | SmolVLM2-2.2B | 51.1 | |
| 9 | LLaVA-OneVision-0.5B | 49.2 | |
| 10 | VILA1.5-3B | 49.1 |
Interactive version: theaggregate.ai/benchmark?slug=cambrian-s-release-percept-test · How It Works · Data refreshed daily, snapshot 2026-10-17.