YOLO11
Ultralytics' 2024 YOLO, five sizes from n to x on COCO's 80 classes, AGPL-3.0 or a paid Enterprise license.
Conditional for size n: code AGPL-3.0-only, weights AGPL-3.0-only. Measured on our RTX 4090: 14.0 ms median, 572 MiB VRAM peak. We scored it 39.1 box AP on COCO val2017. PyTorch weights, 5 MB.
Conditional for size s: code AGPL-3.0-only, weights AGPL-3.0-only. Measured on our RTX 4090: 13.6 ms median, 632 MiB VRAM peak. We scored it 46.1 box AP on COCO val2017. PyTorch weights, 18 MB.
Conditional for size m: code AGPL-3.0-only, weights AGPL-3.0-only. Measured on our RTX 4090: 13.1 ms median, 710 MiB VRAM peak. We scored it 51.0 box AP on COCO val2017. PyTorch weights, 39 MB.
Conditional for size l: code AGPL-3.0-only, weights AGPL-3.0-only. Measured on our RTX 4090: 17.8 ms median, 746 MiB VRAM peak. We scored it 52.7 box AP on COCO val2017. PyTorch weights, 49 MB.
Conditional for size x: code AGPL-3.0-only, weights AGPL-3.0-only. Measured on our RTX 4090: 17.3 ms median, 1.0 GiB VRAM peak. We scored it 54.1 box AP on COCO val2017. PyTorch weights, 109 MB.
License
Conditional
Conditional
Conditional
Conditional
Conditional
- Code
- AGPL-3.0-onlychecked 3 Nov 2026
- Weights
- AGPL-3.0-onlyall sizes, checked 3 Nov 2026
AGPL-3.0: shipping it in a product means releasing your source, unless you buy an Ultralytics Enterprise license.
AGPL-3.0: shipping it in a product means releasing your source, unless you buy an Ultralytics Enterprise license.
AGPL-3.0: shipping it in a product means releasing your source, unless you buy an Ultralytics Enterprise license.
AGPL-3.0: shipping it in a product means releasing your source, unless you buy an Ultralytics Enterprise license.
AGPL-3.0: shipping it in a product means releasing your source, unless you buy an Ultralytics Enterprise license.
On our RTX 4090, measured
14.0ms median
p90 14.7 ms, fp16, batch 1, 640x448
- VRAM peak
- 572 MiB
- Status
- Runs as is
Environment, 3 Nov 2026
NVIDIA GeForce RTX 4090, driver 580.173.02, CUDA 13.0, torch 2.14.1+cu130, Python 3.12.9, ultralytics==8.4.189. Baseline 801 MiB. 10 warm-up + 100 timed. Scope: predict(): pre-processing, forward pass and post-processing; input file decoded beforehand.
Accuracy
39.1box APmeasured
COCO val2017, box AP, 5,000 images, fp16, pycocotools==2.0.11 · 2 Nov 2026
Paper 39.5 COCO mAP50-95Ultralytics YOLO11 docs, detection performance table (mAPval 50-95, docs/en/models/yolo11.md at commit 4fb6485b)
13.6ms median
p90 14.4 ms, fp16, batch 1, 640x448
- VRAM peak
- 632 MiB
- Status
- Runs as is
Environment, 3 Nov 2026
NVIDIA GeForce RTX 4090, driver 580.173.02, CUDA 13.0, torch 2.14.1+cu130, Python 3.12.9, ultralytics==8.4.189. Baseline 801 MiB. 10 warm-up + 100 timed. Scope: predict(): pre-processing, forward pass and post-processing; input file decoded beforehand.
Accuracy
46.1box APmeasured
COCO val2017, box AP, 5,000 images, fp16, pycocotools==2.0.11 · 2 Nov 2026
Paper 47 COCO mAP50-95Ultralytics YOLO11 docs, detection performance table (mAPval 50-95, docs/en/models/yolo11.md at commit 4fb6485b)
13.1ms median
p90 13.4 ms, fp16, batch 1, 640x448
- VRAM peak
- 710 MiB
- Status
- Runs as is
Environment, 3 Nov 2026
NVIDIA GeForce RTX 4090, driver 580.173.02, CUDA 13.0, torch 2.14.1+cu130, Python 3.12.9, ultralytics==8.4.189. Baseline 801 MiB. 10 warm-up + 100 timed. Scope: predict(): pre-processing, forward pass and post-processing; input file decoded beforehand.
Accuracy
51.0box APmeasured
COCO val2017, box AP, 5,000 images, fp16, pycocotools==2.0.11 · 2 Nov 2026
Paper 51.5 COCO mAP50-95Ultralytics YOLO11 docs, detection performance table (mAPval 50-95, docs/en/models/yolo11.md at commit 4fb6485b)
17.8ms median
p90 18.4 ms, fp16, batch 1, 640x448
- VRAM peak
- 746 MiB
- Status
- Runs as is
Environment, 3 Nov 2026
NVIDIA GeForce RTX 4090, driver 580.173.02, CUDA 13.0, torch 2.14.1+cu130, Python 3.12.9, ultralytics==8.4.189. Baseline 801 MiB. 10 warm-up + 100 timed. Scope: predict(): pre-processing, forward pass and post-processing; input file decoded beforehand.
Accuracy
52.7box APmeasured
COCO val2017, box AP, 5,000 images, fp16, pycocotools==2.0.11 · 2 Nov 2026
Paper 53.4 COCO mAP50-95Ultralytics YOLO11 docs, detection performance table (mAPval 50-95, docs/en/models/yolo11.md at commit 4fb6485b)
17.3ms median
p90 17.8 ms, fp16, batch 1, 640x448
- VRAM peak
- 1.0 GiB
- Status
- Runs as is
Environment, 3 Nov 2026
NVIDIA GeForce RTX 4090, driver 580.173.02, CUDA 13.0, torch 2.14.1+cu130, Python 3.12.9, ultralytics==8.4.189. Baseline 801 MiB. 10 warm-up + 100 timed. Scope: predict(): pre-processing, forward pass and post-processing; input file decoded beforehand.
Accuracy
54.1box APmeasured
COCO val2017, box AP, 5,000 images, fp16, pycocotools==2.0.11 · 2 Nov 2026
Paper 54.7 COCO mAP50-95Ultralytics YOLO11 docs, detection performance table (mAPval 50-95, docs/en/models/yolo11.md at commit 4fb6485b)
Weights
Outputs on the shared test photos
Same photos and one score threshold for every model of this task. Failures are marked.
Sizes
| Size | Latency | VRAM | Accuracy | Weights | Runs | License |
|---|---|---|---|---|---|---|
| n | 14.0 ms640x448 | 572 MiB | 39.1 box AP, measured39.5 COCO mAP50-95 paper | 5 MB | As is | Conditional |
| s | 13.6 ms640x448 | 632 MiB | 46.1 box AP, measured47 COCO mAP50-95 paper | 18 MB | As is | Conditional |
| m | 13.1 ms640x448 | 710 MiB | 51.0 box AP, measured51.5 COCO mAP50-95 paper | 39 MB | As is | Conditional |
| l | 17.8 ms640x448 | 746 MiB | 52.7 box AP, measured53.4 COCO mAP50-95 paper | 49 MB | As is | Conditional |
| x | 17.3 ms640x448 | 1.0 GiB | 54.1 box AP, measured54.7 COCO mAP50-95 paper | 109 MB | As is | Conditional |