# YOLO26-L

> YOLO26-L: one-stage object detector from the yolo26 family. Planned for LibreYOLO (not yet packaged).

- Family: yolo26
- Parameters: 24.8M
- GFLOPs: 86.4
- Default input size: 640px
- Detection: anchor-free · nms-free
- Architecture: one-stage · CSPDarknet backbone
- Weight license: non-permissive
- Paper-reported mAP@50-95: 55%

## Benchmarks (COCO val2017)

| Hardware | Runtime | mAP@50-95 | FPS | Latency (ms) | VRAM (MB) |
|---|---|---|---|---|---|
| NVIDIA Jetson Orin Nano Super 8GB | ONNX Runtime FP32 | 53.8 | 1.2 | 855.2 | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 53.8 | 9.3 | 107.2 | 222 |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 53.8 | 24.3 | 41.2 | 11 |
| Raspberry Pi 5 | ncnn FP32 | 53.8 | 1.6 | 641.4 | - |
| Raspberry Pi 5 | ONNX Runtime FP32 | 53.8 | 0.8 | 1241.6 | - |
| Raspberry Pi 5 | PyTorch FP32 | 53.8 | 0.7 | 1377.3 | - |


Source: https://www.visionanalysis.org/model/yolo26l. Benchmarks produced with LibreYOLO (https://github.com/Libre-YOLO/libreyolo).