DK-VLM VSI-Bench - Object Count: leaderboard
DK-VLM v1; LoRA/geometric attention fine-tunes remain distinct from Qwen2.5VL7BInstruct and Qwen3VL8BInstruct backbones. Frozen VGGT-Omega 1B frontend; official shared six-benchmark GeoThinker evaluation settings. Comparison baseline scores are source-reported, starred GeoThinker values reproduced by authors; no automatic primary-board fusion. Metric: Mean relative accuracy (%). Higher is better. Source: https://arxiv.org/html/2610.22356v1
Metric: Mean relative accuracy (%). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DK-VLM-8B | 75.91 | |
| 2 | Q-GeoMem-7B | 74.8 | |
| 3 | GeoThinker-8B | 73.93 | |
| 4 | Cambrian-S-7B | 73.2 | |
| 5 | DK-VLM-7B | 72.02 | |
| 6 | GeoThinker-7B | 71.96 | |
| 7 | VG-LLM-8B | 71.4 | |
| 8 | SpatialStack-5B | 71 | |
| 9 | VLM-3R-7B | 70.2 | |
| 10 | Qwen3-VL-8B (DK-VLM paper baseline) | 67.6 |
Interactive version: theaggregate.ai/benchmark?slug=dk-vlm-vsi-bench-object-count · How It Works · Data refreshed daily, snapshot 2026-10-17.