Runs On
What it takes to serve sam3 (860M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.
SAVRN's Notes on sam3
Segmentation is the job: a text phrase, a point, a box or a mask, and it returns every instance of that concept in an image or video, tracked across frames. At 16-bit the 860M parameters take 1.7 GB and the run needs 2.1 GB, so on the cheapest Index listing, one MI300X with 192 GB at $1.85 an hour, almost the whole card stays free; share it. The download is 6.9 GB across 12 files, four times the 16-bit weights.
The license field says other, no summary on file, and access is gated, so AI at Meta grants it before any file moves; read the terms before this goes into a product. No context length, base model or dataset is listed, so its lineage is whatever the publisher says. The one third-party number we hold is a 44.4 average on PBench.
SAVRN Research, 2026-09-18
Model Card
SAM 3 is a unified foundation model for promptable segmentation in images and videos. It can detect, segment, and track objects using text or visual prompts such as points, boxes, and masks. Compared to its predecessor SAM 2, SAM 3 introduces the ability to exhaustively segment all instances of an open-vocabulary concept specified by a short text phrase or exemplars. Unlike prior work, SAM 3 can handle a vastly larger set of open-vocabulary prompts. It achieves 75-80% of human performance on our new SA-CO benchmark which contains 270K unique concepts, over 50 times more than existing benchmarks. The official code is publicly released in the sam3 repo. SAM3 performs Promptable Concept…
Excerpt from the card by AI at Meta, licensed other.
Identity and Version
- Repository
- facebook/sam3
- Publisher
- AI at Meta
- Task
- Mask generation
- Modality
- Other
- Library
- transformers
- Parameters
- 860M parameters
- Languages
- en
- Revision
- 3c879f39826c281e95690f02c7821c4de09afae7
- First published
- 2025-11-07
- Last updated
- 2025-11-20
Files and Weights
12 files, 6.9 GB in total. The weights are 2 files totalling 6.9 GB in pt, safetensors.
Weights2 files · 6.9 GB
Configuration3 files · 28.1 KB
Tokenizer4 files · 5.0 MB
Documentation2 files · 33.4 KB
Repository1 file · 1.5 KB
Every file
License and Download
- License
- other
- Access
- Access requested at publisher
- Download size
- 6.9 GB
Evaluations
Each result is shown as reported, with the conditions its reporter stated. None is a SAVRN measurement. A comparison lines two results up only when their configuration, unit and setup are all stated and identical.
Memory Requirements
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About sam3
How much GPU memory does sam3 need?
About 2.1 GB at 16-bit and 0.5 GB at 4-bit: the weights (860M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run sam3 on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
What license is sam3 released under?
other, as its publisher declares it. Read the license text before commercial use.
Similar Models
If you want to give the Finegrain Box Segmenter a try, the best way to is take a look at the Finegrain Object Cutter Space we shipped on Hugging Face: it's a fun "prompt to cut out" experience that will enable you to create pixel quality and high resolution cutouts for any object in a photo, by just naming the object. While building Finegrain, we needed a way to create pixel perfect and high resolution cutouts for objects in images. We looked at off-the-shelf solutions, but they simply didn't work for us: - On the one hand, traditional background removal models are great at producing HD cutouts, but unfortunately, different people will have different definitions for background and…
Open weights
mit
95M parameters
refiners
