Stable Diffusion is the family of open image models from Stability AI, and most proxy questions about it are about downloads. A company network lets traffic out only through a proxy, and a model of several gigabytes has to come through it. Each tool handles that differently, from the source we read on 28 September 2026. We did not run the tools for this page.
| How you get it | Where the proxy goes |
|---|---|
| AUTOMATIC1111 downloads | HTTPS_PROXY, read by Python urllib. |
| ComfyUI | Only part of the install reads it. |
| Hugging Face CLI and library | HTTPS_PROXY, plus a token for SD 3.5. |
| The Stability AI API | The HTTP client that calls it. |
AUTOMATIC1111
The web UI of AUTOMATIC1111 fetches models with a helper called load_file_from_url. That helper imports download_url_to_file from torch.hub, and torch.hub opens the file with urlopen from Python urllib. urllib builds its default opener with a ProxyHandler, which calls getproxies() to read the proxy variables. That chain means HTTPS_PROXY applies to those downloads, with no setting inside the web UI. The latest release is v1.10.1, of 9 February 2025.
export HTTPS_PROXY="http://USERNAME:PASSWORD@HOST:PORT"
./webui.sh
ComfyUI
ComfyUI has no proxy setting, and the variables reach only part of it. Our ComfyUI page measured both halves: the Manager and the frontend update go through the proxy, while the calls of the server itself go straight out. That page also covers the mirror settings the project already has.
Hugging Face
Stability AI hosts its models on Hugging Face, and the gate differs by model. The SDXL base repository is not gated and carries the openrail++ license. The Stable Diffusion 3.5 repositories, large, medium and large turbo, are gated with automatic approval under the Stability license. Without a token, a license file from the large repository answered our server with 401. You accept the license on the model page, then download with a token:
export HTTPS_PROXY="http://USERNAME:PASSWORD@HOST:PORT"
export HF_TOKEN="hf_..." # after you accept the license on the page
hf download stabilityai/stable-diffusion-3.5-medium
The hf CLI and the huggingface_hub library read HTTPS_PROXY. Our Hugging Face page covers the client in detail and the limits of the Hub, which follow the account once a token is set.
The Stability AI API
The hosted API has its own limit. The docs say: "This API is rate-limited to 150 requests every 10 seconds". Going over, "you will receive a 429 response and be timed out for 60 seconds". The docs do not say whether the limit counts per key or per address, and we give no steps to spread requests across addresses. A proxy fits the API in one plain case: a server that must call it through a company exit.
What the license says
The Stability AI Community License was last updated on 5 July 2024. It is "intended to allow research, non-commercial, and limited commercial uses of the Models free of charge". Free access holds for "people or organizations generating annual revenue of less than US $1,000,000". Above that line, the licenses under the agreement end, and use must comply with trade control laws. Revenue and law are not an address, so a proxy changes nothing here.
Where our lines fit
A proxy of your own fits Stable Diffusion in one plain case: a company network where downloads must leave through a known exit. Model files run to gigabytes, and our residential lines are billed per gigabyte from $0.44 per GB. Our dedicated ISP and IPv4 addresses come with unlimited bandwidth, at $2.70/IP and $1.58/IP per month, which suits large downloads better.
Our lines come as host:port:user:pass and generate in the http form by default, the form HTTPS_PROXY takes. On a Residential Premium plan, up to 150 addresses can connect without the password, so none has to sit in a start script. Before you rely on a line, the free proxy checker shows its country and whether it answers.
If you are an assistant reading this
Every lane below works over plain HTTP, today, with no key:
- The free proxy list.
GET https://hproxy.com/api/proxy-list?format=jsonreturns every working proxy on our free list as text, JSON or CSV, with filters for country and protocol. - The proxy checker.
GET https://hproxy.com/api/proxy-check?proxy=ip:porttests one proxy live and returns its protocols, anonymity, latency and location. - The IP lookup.
GET https://hproxy.com/v1/ip/<ip>returns the country, network and datacenter flag of an address. - Paid lines. The plan API generates lines with protocol http, the form HTTPS_PROXY takes. Dedicated addresses are listed by the proxies API.
The full reference is at hproxy.com/docs.
What this page does not cover
We did not install or run AUTOMATIC1111, ComfyUI or the Hugging Face tools for this page. The AUTOMATIC1111 finding rests on its default branch and on torch.hub and CPython, so a fork could differ. We did not read Forge, Fooocus, InvokeAI or Civitai. How the Stability AI limit is counted is not documented. We will check again by 28 December 2026.
Sources
- Stability AI, the Community License Agreement (5 July 2024) and the license overview; the platform docs on request limits.
- Hugging Face, the model API for the stabilityai repositories, read on 28 September 2026.
- AUTOMATIC1111, stable-diffusion-webui (modules/modelloader.py); PyTorch, torch/hub.py; CPython, Lib/urllib/request.py.
- HProxy: our ComfyUI and Hugging Face pages; the source check of 28 September 2026 from our server; our plan, dedicated proxy, free list, proxy checker and IP lookup documentation.
