FindIt - Instance Detection: leaderboard
Instance detection: the model finds the specific object shown in a reference image within a scene, on HR-InsDet (easy and hard) and RoboTools. FindIt (2026), a format-informed localization benchmark for multimodal LLMs, scores each model with the bounding-box and output format it handles best, as F1 at IoU 0.5 averaged over the datasets, in percent; higher is better.
Metric: Average F1@0.5 (%) over instance detection (find the specific object shown in a reference image) on HR-InsDet easy and hard and RoboTools, 1,000 queries each; each model scored with the box representation, text or JSON output and JSON key that a two-stage probe on Pascal VOC found best for it; boxes Hungarian-matched to the ground truth, a match counts at IoU 0.5 or more (with the right label for multi-label queries); unparseable outputs count as empty; higher is better. Source: arxiv.org. Saturation forecast: Around January 2028. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 9B (Thinking) | 34.8 | #803 (Qwen 3.5 9B) |
| 2 | Qwen 3.5 9B (Non-reasoning) | 34.5 | #803 (Qwen 3.5 9B) |
| 3 | Qwen 3 VL 8B Instruct | 27.7 | #981 |
| 4 | Gemini 2.5 Flash (Non-reasoning) | 18 | #453 (Gemini 2.5 Flash) |
| 5 | GPT-5.4 (Non-reasoning) | 16.1 | #123 (GPT-5.4) |
| 6 | Qwen 2.5 VL 7B Instruct | 5.7 | #1833 |
| 7 | InternVL3-8B | 0.5 | #1688 |
| 8 | Gemma 4 E4B | 0.3 | #2214 |
| 9 | Claude Sonnet 4.5 | 0 | #250 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=findit-instance-detection · How It Works · Data refreshed daily, snapshot 2026-11-13.