YOLO11-seg
Ultralytics' 2024 YOLO for instance segmentation, 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: 14.5 ms median, 596 MiB VRAM peak. We scored it 31.8 mask AP on COCO val2017. PyTorch weights, 6 MB.
Conditional for size s: code AGPL-3.0-only, weights AGPL-3.0-only. Measured on our RTX 4090: 14.7 ms median, 640 MiB VRAM peak. We scored it 37.3 mask AP on COCO val2017. PyTorch weights, 20 MB.
Conditional for size m: code AGPL-3.0-only, weights AGPL-3.0-only. Measured on our RTX 4090: 14.5 ms median, 756 MiB VRAM peak. We scored it 41.2 mask AP on COCO val2017. PyTorch weights, 43 MB.
Conditional for size l: code AGPL-3.0-only, weights AGPL-3.0-only. Measured on our RTX 4090: 18.8 ms median, 792 MiB VRAM peak. We scored it 42.4 mask AP on COCO val2017. PyTorch weights, 53 MB.
Conditional for size x: code AGPL-3.0-only, weights AGPL-3.0-only. Measured on our RTX 4090: 19.6 ms median, 1.1 GiB VRAM peak. We scored it 43.4 mask AP on COCO val2017. PyTorch weights, 119 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
14.5ms median
p90 14.9 ms, fp16, batch 1, 640x448
- VRAM peak
- 596 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
31.8mask APmeasured
COCO val2017, mask AP, 5,000 images, fp16, pycocotools==2.0.11 · 2 Nov 2026
Paper 32 COCO mAPmask 50-95Ultralytics YOLO11 docs, segmentation performance table (mAPmask 50-95, docs/en/models/yolo11.md at commit 4fb6485b)
14.7ms median
p90 14.9 ms, fp16, batch 1, 640x448
- VRAM peak
- 640 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
37.3mask APmeasured
COCO val2017, mask AP, 5,000 images, fp16, pycocotools==2.0.11 · 2 Nov 2026
Paper 37.8 COCO mAPmask 50-95Ultralytics YOLO11 docs, segmentation performance table (mAPmask 50-95, docs/en/models/yolo11.md at commit 4fb6485b)
14.5ms median
p90 14.9 ms, fp16, batch 1, 640x448
- VRAM peak
- 756 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
41.2mask APmeasured
COCO val2017, mask AP, 5,000 images, fp16, pycocotools==2.0.11 · 2 Nov 2026
Paper 41.5 COCO mAPmask 50-95Ultralytics YOLO11 docs, segmentation performance table (mAPmask 50-95, docs/en/models/yolo11.md at commit 4fb6485b)
18.8ms median
p90 19.7 ms, fp16, batch 1, 640x448
- VRAM peak
- 792 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
42.4mask APmeasured
COCO val2017, mask AP, 5,000 images, fp16, pycocotools==2.0.11 · 2 Nov 2026
Paper 42.9 COCO mAPmask 50-95Ultralytics YOLO11 docs, segmentation performance table (mAPmask 50-95, docs/en/models/yolo11.md at commit 4fb6485b)
19.6ms median
p90 20.0 ms, fp16, batch 1, 640x448
- VRAM peak
- 1.1 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
43.4mask APmeasured
COCO val2017, mask AP, 5,000 images, fp16, pycocotools==2.0.11 · 2 Nov 2026
Paper 43.8 COCO mAPmask 50-95Ultralytics YOLO11 docs, segmentation performance table (mAPmask 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 | 14.5 ms640x448 | 596 MiB | 31.8 mask AP, measured32 COCO mAPmask 50-95 paper | 6 MB | As is | Conditional |
| s | 14.7 ms640x448 | 640 MiB | 37.3 mask AP, measured37.8 COCO mAPmask 50-95 paper | 20 MB | As is | Conditional |
| m | 14.5 ms640x448 | 756 MiB | 41.2 mask AP, measured41.5 COCO mAPmask 50-95 paper | 43 MB | As is | Conditional |
| l | 18.8 ms640x448 | 792 MiB | 42.4 mask AP, measured42.9 COCO mAPmask 50-95 paper | 53 MB | As is | Conditional |
| x | 19.6 ms640x448 | 1.1 GiB | 43.4 mask AP, measured43.8 COCO mAPmask 50-95 paper | 119 MB | As is | Conditional |