Model weights · FastSAM
Download FastSAM weights
A CNN take on segment-anything built on YOLOv8-seg, reaching similar masks far faster than the original transformer-based SAM. Pick a size below and download the official .pt checkpoint in one click, with published accuracy, parameters and license all in view.
FastSAM checkpoints, pick a size and download
A CNN take on segment-anything built on YOLOv8-seg, reaching similar masks far faster than the original transformer-based SAM.
| Model | Params | Download |
|---|---|---|
FastSAM-s FastSAM-s.pt | 11.8M | |
FastSAM-x FastSAM-x.pt | 72.2M |
Figures are published by the authors. Weights host: github.com. Clicking Download verifies the file and starts it straight from the official CDN.
How to load FastSAM weights
Install the Ultralytics package, then point the loader at the checkpoint. It downloads automatically on first use, or you can pass the local path to the file you downloaded above.
pip install ultralyticsfrom ultralytics import FastSAM
# Downloads on first use, or pass a local path to your .pt file
model = FastSAM("FastSAM-s.pt")
results = model("image.jpg")About FastSAM
FastSAM reframes segment-anything as a YOLOv8-seg instance-segmentation problem plus prompt-guided selection, reaching comparable masks at up to 50x the speed of the original SAM. A pragmatic CNN alternative when you want SAM-style prompting without a transformer's cost.
- Author
- CASIA IVA
- Released
- 2023
- Tasks
- Segment
- Framework
- Ultralytics (PyTorch)
- Input size
- 1024px
- License
- AGPL-3.0
More model weights to download
Each page lists every checkpoint with accuracy, parameters, license and a one-click download.
From the blog
Tutorials, code, and notes on computer vision, deep learning, and applied AI.
Build a semantic image search engine with CLIP and Python
Learn how to build a semantic image search engine that finds pictures by meaning. A few lines of Python turn a folder of images into a searchable index you can query in plain English.
YOLO26 vs YOLO11: Real-time ONNX FPS benchmark in Python
Build one small, reusable class that runs Ultralytics YOLO26 and Ultralytics YOLO11 as ONNX models, draws clean detections, and overlays live FPS and latency so you can compare their real-time speed on the exact same footage.
Depth estimation from a single image with Depth Anything V2
Learn how to estimate depth from a single image, a video, or your webcam using Depth Anything V2 and a clean, reusable Python class.
Ultralytics object trackers comparison: ByteTrack, BoT-SORT & More
How do the six Ultralytics trackers actually behave on the same footage? A look at BoT-SORT, ByteTrack, OC-SORT, Deep OC-SORT, FastTrack, and TrackTrack, their internals, trade-offs, and side-by-side results on ID switches, ID stability, and FPS.
How to extract text from images with LightOnOCR and Python
Learn how to read text from images using LightOnOCR, a small and fast vision language model, with a clean and reusable Python class.
Object tracking and trajectory forecasting with YOLO26 and ByteTrack
Detect, track, and predict the future path of people and vehicles using Ultralytics YOLO26, ByteTrack, and a lightweight velocity-based forecasting model.
Real time bird detection and tracking using YOLO11
Learn how to detect and track birds using Ultralytics YOLO11 for real-time monitoring and ecological research through computer vision.
How to count people in zones with YOLO26 and OpenCV
A practical walkthrough of a compact Python script that detects, tracks, and counts people inside polygon zones using Ultralytics YOLO26 and OpenCV.
Build a semantic image search engine with CLIP and Python
Learn how to build a semantic image search engine that finds pictures by meaning. A few lines of Python turn a folder of images into a searchable index you can query in plain English.
YOLO26 vs YOLO11: Real-time ONNX FPS benchmark in Python
Build one small, reusable class that runs Ultralytics YOLO26 and Ultralytics YOLO11 as ONNX models, draws clean detections, and overlays live FPS and latency so you can compare their real-time speed on the exact same footage.
Depth estimation from a single image with Depth Anything V2
Learn how to estimate depth from a single image, a video, or your webcam using Depth Anything V2 and a clean, reusable Python class.
Ultralytics object trackers comparison: ByteTrack, BoT-SORT & More
How do the six Ultralytics trackers actually behave on the same footage? A look at BoT-SORT, ByteTrack, OC-SORT, Deep OC-SORT, FastTrack, and TrackTrack, their internals, trade-offs, and side-by-side results on ID switches, ID stability, and FPS.
How to extract text from images with LightOnOCR and Python
Learn how to read text from images using LightOnOCR, a small and fast vision language model, with a clean and reusable Python class.
Object tracking and trajectory forecasting with YOLO26 and ByteTrack
Detect, track, and predict the future path of people and vehicles using Ultralytics YOLO26, ByteTrack, and a lightweight velocity-based forecasting model.
Real time bird detection and tracking using YOLO11
Learn how to detect and track birds using Ultralytics YOLO11 for real-time monitoring and ecological research through computer vision.
How to count people in zones with YOLO26 and OpenCV
A practical walkthrough of a compact Python script that detects, tracks, and counts people inside polygon zones using Ultralytics YOLO26 and OpenCV.
Frequently asked questions
- How do I download FastSAM weights?
- Click the Download button next to any variant in the table above; we verify the file and start it straight from the official CDN. You can also let the loader fetch it automatically on first use with `model = FastSAM("FastSAM-s.pt")`. All 2 FastSAM checkpoints are hosted officially.
- Is FastSAM free for commercial use?
- FastSAM is released under AGPL-3.0. That is free for open-source and research; closed-source commercial deployments need an Ultralytics Enterprise license. Always confirm against the linked license text.
- Which FastSAM model size should I use?
- Start with FastSAM-s, the smallest and fastest, ideal for prototyping, edge and CPU. Move up the ladder only when you need more accuracy and have the compute for it.