Is Hugging Face Down?

No — Hugging Face is up

Reachable from all 8 checked regions

Average response time: 363ms

Last checked 10 hours ago · checks run every 6 hours

Official status page: https://status.huggingface.co

Hugging Face uptime

100%

Last 7 days

100%

Last 30 days

100%

Last 90 days

310ms

Avg response, 30 days

Measured from multiple regions every 6 hours. Percentages count only checks that returned an availability answer — 13 days measured so far. A dash means that window does not yet hold enough measured days to publish a figure.

30-day history

13-day clean streak

Jul 25 Today

No downtime Partial Downtime Not measurable No data

Reachability by region

Each region runs its own request from a different part of the world. A service can be up for one continent and down for another, which is usually the first sign of a routing or CDN problem.

gru

401ms

DNS 1ms TCP 4ms TLS 6ms TTFB 178ms

iad

68ms

DNS 0ms TCP 1ms TLS 4ms TTFB 57ms

jnb

847ms

DNS 0ms TCP 0ms TLS 3ms TTFB 317ms

lax

277ms

DNS 0ms TCP 2ms TLS 5ms TTFB 119ms

lhr

366ms

DNS 0ms TCP 2ms TLS 4ms TTFB 141ms

nrt

317ms

DNS 0ms TCP 1ms TLS 4ms TTFB 239ms

ord

108ms

DNS 0ms TCP 2ms TLS 3ms TTFB 83ms

sin

788ms

DNS 1ms TCP 1ms TLS 3ms TTFB 299ms

sjc

135ms

DNS 0ms TCP 2ms TLS 4ms TTFB 123ms

What Hugging Face does

Hugging Face hosts machine learning models, datasets and Spaces demos, and serves them through the Hub and its Inference Endpoints. Libraries such as transformers download model weights from it at runtime, so training jobs, notebooks and deployed applications that pull a model by name depend on it even when nothing in the code names the site directly.

What an outage looks like

The status page separates the Hub, git hosting and serving, the Inference Endpoints UI and API, Spaces and the Spaces proxy, plus its AWS and Google Cloud CDN components. Downloads stall or return 5xx partway through large weight files, calls that fetch a model by name hang, Spaces show a build or runtime error, and pushes to a model repository are rejected.

What to do about it

Check status.huggingface.co, which reports the Hub, Inference Endpoints and the CDNs separately, since a CDN fault breaks downloads while the website itself loads normally. Models already in your local cache keep working without the Hub, and the libraries have an offline mode that stops them reaching out mid-run. Clear any partially downloaded weights before retrying, since a truncated file fails in confusing ways.

Is it down for everyone, or just you?

If this page says Hugging Face is up but it is not loading for you, the problem is between you and them. Run a check against any URL from all 18 regions to find out where it breaks.

Test it yourself

Related services

Hugging Face outage FAQ

Why is my model download stuck at the same percentage?

Large weight files are served through CDN components that the status page tracks separately from the Hub itself, so downloads can stall while the website stays responsive. A partial file left in the cache will then fail on the next attempt in ways that look like a corrupt model rather than a network problem. Delete the incomplete download before retrying, and check the CDN lines specifically.

Can I run my code while Hugging Face is down?

If the model is already in your local cache, yes. The libraries look there before reaching the network, and they provide an offline mode that prevents any Hub calls at all, which is worth setting during an incident so a background metadata request does not stall your job. Anything you have never downloaded on that machine will not be available until the Hub returns.

Does a Hub outage stop my Inference Endpoint?

Not necessarily. Inference Endpoints are tracked as their own components, separate from the Hub and from Spaces, so a deployed endpoint can keep serving while the website and model downloads are degraded. The reverse also happens. Check the specific Inference Endpoints lines rather than the overall banner, which covers a platform much broader than the part your application calls.

Why did my Space stop working?

Spaces and the Spaces proxy are separate components, and a Space can fail either because the platform is degraded or because it is rebuilding and pulling a model or dataset that is currently unreachable. During a Hub incident a rebuild is the riskiest moment, since it re-downloads dependencies. If the status page is clean, look at the Space build logs before assuming a platform fault.

How we measure this

We request Hugging Face's public endpoint every 6 hours from Fly.io regions across six continents — 8 of them answered the most recent check.
A region counts as down only when it gets no usable HTTP response. A 403 or 429 means the origin answered and refused us, which we report as blocked, never as an outage.
A single failing region is treated as probe noise. We only change the verdict when two consecutive cycles agree.
Response times average only the regions that actually served the page, so a timeout never inflates the number.
Where Hugging Face publishes an official status feed we read it too. An all-clear from the vendor can soften an unconfirmed degradation; a vendor-declared outage only worsens our verdict when our own checks corroborate it.

Get alerted when Hugging Face goes down

This page refreshes every 6 hours. Your own monitors run as often as every 30 seconds, from the same 18 regions, and tell you the moment something breaks.

Start Free Monitoring

Free plan available No credit card required