Hugging Face est-il en panne ?
Modèles, jeux de données et inférence.
Disponibilité sur 90 jours
Historique des incidents
Scheduled maintenance: Hub
Maintenance completed
Scheduled maintenance: Hub
During a brief window (approximately 5–10 minutes), uploads and commits to repositories may temporarily fail or need to be retried while we upgrade database infrastructure. This affects both the UI and the API. Downloads, reads, and browsing are not affected, and no action is needed on your end.
Scheduled maintenance: Spaces
During a brief window (approximately 5–10 minutes), write and management operations on Spaces (creating or duplicating a Space, changing hardware or settings, pausing/restarting, and updating secrets and variables) may temporarily return a 503 error while we upgrade infrastructure. This affects both the UI and the API. Reads and already-running Spaces are not affected, and no action is needed on y
Elevated error rate - AWS CDN (Singapore)
Our fix is now in place and this incident is resolved. Applying it caused one final short period of failed downloads between 23:00 and 23:14 UTC, as we noted could happen. Downloads from the Singapore region have been stable since, and we have added capacity to prevent this from recurring. We are sorry for the disruption.
Elevated error rate - AWS CDN (Singapore)
Some file downloads from our Singapore region (ap-southeast-1) failed with 500 errors during two short periods, starting at 15:14 UTC. Downloads are normal again now. The cause is not fully resolved, so short bursts of errors can return until our fix is in place. Other CDN regions are not affected. We are sorry for the disruption.
Scheduled maintenance: Inference Endpoints
During a short window within this period (roughly 5–10 minutes), endpoint management operations (creating, updating, pausing, or resuming endpoints) may briefly return a 503 ("temporarily unavailable due to maintenance") while we complete infrastructure upgrades. Running endpoints and inference requests are not affected, and no action is needed on your end.
Scheduled maintenance: Inference Endpoints
During a short window inside this period (roughly 5–10 minutes) endpoint management operations (creating, updating, pausing, or resuming endpoints) may briefly return a 503 ("temporarily unavailable due to maintenance") while we complete upgrades to infrastructure. Running endpoints and inference requests are not affected, and no action is needed on your end.
Hub unavailable
Services are back to nominal operations. Root cause confirmed to global Amazon Web Services outage.
Hub unavailable
We are mitigating the issue, and services are coming back up.
Hub unavailable
Hugging Face platform is unavailable from most regions in the world due following a global outage from Amazon Web Services. We are working on a mitigation.
Scheduled maintenance: Jobs
During a short window inside this period (roughly 5–10 minutes) write operations may briefly return a 503 ("temporarily unavailable due to maintenance") while we complete upgrades to infrastructure. Reads and already-running jobs are not affected, and no action is needed on your end.
Elevated error rate – AWS CDN (Singapore)
Resolved. Error rates returned to normal at 08:05 UTC and the CDN is operating normally. The disruption lasted approximately 7 minutes. We apologize for any inconvenience.
Elevated error rate – AWS CDN (Singapore)
Between 07:58 and 08:05 UTC, the AWS CDN returned an elevated rate of HTTP 500 errors for a portion of requests in the Asia-Pacific (Singapore / ap-southeast-1) region. The cause was a transient internal service disruption, which has been mitigated.
Data serving slowness in Asia region
Issue is solved, Asia traffic back to nominal bandwidth performances.
Data serving slowness in Asia region
We are experiencing slowness on large file downloads in Asia region since 7AM UTC. This only affects downloads via the huggingface CLI. Root cause identified and mitigation in progress.
Source officielle
Les données de cette page proviennent exclusivement de la page de statut publique de Hugging Face.
Page de statut officielleQuestions fréquentes
Hugging Face est-il en panne actuellement ?
Au dernier relevé (2026-09-14 09:49 UTC), Hugging Face est en état « Opérationnel » d'après sa page de statut officielle. 11 incident(s) en cours : Scheduled maintenance: Hub.
Quelle est la disponibilité de Hugging Face sur 90 jours ?
L'historique de Hugging Face est en cours de constitution sur Upstream Status.
Comment être prévenu automatiquement d'une panne de Hugging Face ?
Créez une alerte gratuite sur Upstream Status : dès qu'un incident est publié sur la status page de Hugging Face, vous recevez un email (ou une notification Slack/webhook avec un plan payant).