How much GPU memory does depth-anything-v2-small need?

About 1.1 GB: 0.1 GB of weights and the runtime's own memory.

What is the cheapest GPU to run depth-anything-v2-small?

Right now one GTX 1080 at $0.08 an hour; RTX 5090 at $0.69 an hour is the best value for its speed. Prices are live offers from GPU clouds, checked every 15 minutes.

Can depth-anything-v2-small run on an RTX 4090?

Yes: it needs about 1.1 GB and an RTX 4090 has 24 GB.

Models / Depth-Anything-V2-Small

onnx-community/depth-anything-v2-small

Depth estimation · served with ONNX Runtime · ↓ 119.9K downloads a month · View on Hugging Face

onnx-community/depth-anything-v2-small needs about 1.1 GB of GPU memory. It runs on one GTX 1080 from $0.08/h right now; best value: RTX 5090 at $0.69/h.

Weights: 100 MB
GPU memory needed: about 1.1 GB
Memory · disk: 8 GB · 30 GB
Licence: apache-2.0

Versions

Smaller versions (8-bit, 4-bit) need less GPU memory and answer a little less precisely.

full precisiononnx/model.onnx100 MB · 1.1 GB GPU 4-bit (bnb)onnx/model_bnb4.onnx30 MB · 1 GB GPU float16onnx/model_fp16.onnx50 MB · 1.1 GB GPU 8-bitonnx/model_int8.onnx30 MB · 1 GB GPU 4-bitonnx/model_q4.onnx30 MB · 1 GB GPU 4-bit, float16onnx/model_q4f16.onnx20 MB · 1 GB GPU 8-bitonnx/model_quantized.onnx30 MB · 1 GB GPU 8-bitonnx/model_uint8.onnx30 MB · 1 GB GPU

Hardware and deploy

Deploying ONNX models from the catalog is coming soon. Today, download the files from Hugging Face and upload them in Inference.

Questions

How much GPU memory does depth-anything-v2-small need?: About 1.1 GB: 0.1 GB of weights and the runtime's own memory.
What is the cheapest GPU to run depth-anything-v2-small?: Right now one GTX 1080 at $0.08 an hour; RTX 5090 at $0.69 an hour is the best value for its speed. Prices are live offers from GPU clouds, checked every 15 minutes.
Can depth-anything-v2-small run on an RTX 4090?: Yes: it needs about 1.1 GB and an RTX 4090 has 24 GB.

Loading…