SAM3-GGUF

GGUF weights for SAM 3 (Segment Anything Model 3), converted for use with sam3.cpp.

Available Variants

File	Precision	Size	Inference (M4 Max)
sam3-image-f32.gguf	F32	3.2 GB	3.5s
sam3-image-f16.gguf	F16	1.7 GB	3.5s
sam3-image-q8_0.gguf	Q8_0	1.0 GB	3.2s

Usage

import sam3cpp

model = sam3cpp.Sam3Model("hf://rob-laz/sam3-gguf/sam3-image-f16.gguf")
pred = model.predict("photo.jpg", "person")
print(f"{pred.count} detections")

See sam3.cpp for build instructions and full documentation.

Conversion

Converted from the original SAM 3 MLX weights using tools/convert_mlx_sam3_to_gguf.py.