LLM Stats (ODinW): leaderboard
Object Detection in the Wild — 35 datasets of diverse real-world object detection tasks spanning aerial imagery, medical scans, and everyday scenes.
Metric: Score (%). Source: llm-stats.com. Saturation forecast: Around 2030. 16 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.6 Plus | 51.8 | #120 |
| 2 | Qwen 3.7 Plus | 51.1 | #92 |
| 3 | Qwen 3.6 35B A3B | 50.8 | #226 |
| 4 | Qwen 3 VL 235B A22B Instruct | 48.6 | #273 |
| 5 | Qwen 3 VL 4B Instruct | 48.2 | #515 |
| 6 | Qwen 3 VL 30B A3B Instruct | 47.5 | #384 |
| 7 | Qwen 3 VL 32B Instruct | 46.6 | #283 |
| 8 | Qwen 3 VL 8B Instruct | 44.7 | #420 |
| 9 | Qwen 3.5 122B A10B | 44.5 | #173 |
| 10 | Qwen 3 VL 235B A22B (Thinking) | 43.2 | #238 (Qwen 3 VL 235B A22B) |
| 11 | Qwen 3.5 35B A3B | 42.6 | #256 |
| 12 | Qwen2.5-Omni-7B | 42.4 | #624 |
| 13 | Qwen 3 VL 30B A3B (Thinking) | 42.3 | #351 (Qwen 3 VL 30B A3B) |
| 14 | Qwen 3.5 27B | 41.1 | #187 |
| 15 | Qwen 3 VL 8B (Thinking) | 39.8 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=llm-stats-odinw · How It Works · Data refreshed daily, snapshot 2026-10-20.