Roboflow Vision Evals - Object Detection (mAP@50)
Multimodal · 2026-07-22
The Roboflow Vision Evals object-detection task measures localization and classification performance on 250 samples. Its headline score is mean average precision at an intersection-over-union threshold of 0.50 (mAP@50).
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 3.5 Flash | 68.7 |
| GPT-5.6 Sol | 68.2 |
| Gemini 3.1 Pro Preview | 67.4 |
| GPT-5.6 Terra | 60.7 |
| Qwen3.7-Plus | 60.1 |
| GPT-5.6 Luna | 59.9 |
| Qwen3.5-27B | 58.8 |
| Muse Spark 1.1 | 58.2 |
| Gemini 3.5 Flash-Lite | 57.5 |
| GLM 5V Turbo | 56.1 |
| Claude Fable 5 | 56.0 |
| Gemini 3.6 Flash | 56.0 |
| Qwen3-VL-235B-A22B-Instruct | 52.2 |
| Kimi K3 | 51.9 |
| GPT-5.5 | 41.7 |
| Gemini 3 Flash Preview | 38.6 |
| Opus 4.8 | 38.6 |
| Sonnet 5 | 36.1 |
| Kimi K2.6 | 35.7 |
| Gemini 2.5 Pro | 33.7 |
| GPT-5.4 Mini | 16.1 |