OpenAI is investigating a complete failure affecting its Workspace Agent service, preventing users from initiating new runs. The status page lists the component as experiencing a full outage, meaning new executions cannot be started during this window.The service is down, so you cannot start any new automated tasks right now.The Workspace Agent interface is rejecting new initiation requests due to a full outage.
The figures
The incident page identifies one affected component: the Agent service. The severity is described as a full outage, indicating total unavailability for new tasks. No numerical metrics on affected user counts or duration are provided in the current update.
Behind the figures
These details come from an official advisory posted on the OpenAI status dashboard. The entry provides a summary stating the team is investigating the issue preventing new starts. No internal metrics or root cause analysis have been shared publicly at this time.
What it means for your stack
Developers relying on the Workspace Agent face immediate disruption for new tasks. New tasks cannot be queued or executed, which halts any pipelines dependent on starting fresh runs. The inability to start new ones blocks fresh inference requests via this specific interface.
Building on this
Inference engineers must account for total unavailability of the agent interface when designing systems. Since the outage prevents new starts, systems relying on this interface will fail to initiate tasks. The documented limit is a hard stop on initiation, requiring external monitoring of the status page for restoration signals.
Why it matters
- Developers cannot queue new inference tasks, stopping workflows that depend on starting new agent runs.
- The full outage status indicates the Agent component is unavailable for new initiations.
- Lack of a confirmed fix timeline forces teams to monitor the status page manually for resolution updates.
- This highlights the dependency risk of relying on a single agent interface for application logic.
Topics: OpenAI, Workspace Agent, outage, inference, status
Updated
Then / now / next
The Workspace Agent service was operational, allowing users to start new runs.
A full outage prevents any new agent runs from starting, with the status listed as investigating.
The next update depends on OpenAI's investigation findings, with no scheduled resolution time currently published.
Story timeline
- Then The Workspace Agent service was operational, allowing users to start new runs.
- 11 Nov 2026 Unable to start new Workspace Agent runs
- Now A full outage prevents any new agent runs from starting, with the status listed as investigating.
- Next The next update depends on OpenAI's investigation findings, with no scheduled resolution time currently published.
Did this article give you the evidence you needed?
Your choice stays on this device. No public tally.Terms in this story
Workspace Agent
An OpenAI service component for executing automated tasks or workflows.
Full outage
A state where a service is completely unavailable for its intended function.
Status page
A public dashboard used by vendors to report service health and incidents.
Questions readers ask
Can I still run existing Workspace Agent tasks?
The advisory specifically mentions preventing new runs; it does not state the status of existing runs.
Is there a workaround for starting new agent runs?
No workaround is provided in the source material; the service is experiencing a full outage for new initiations.
When will the service be restored?
OpenAI has not provided a timeline, stating only that they are investigating and will share updates as information becomes available.
Connected stories
Reading receipt
- OpenAI reports a full outage of the Workspace Agent component.
- Users are currently unable to start new runs via the service.
- The status page lists the issue as investigating with no resolution time.
Still open: What is the root cause of the failure preventing new runs?
Still open: How long will the outage last before service is restored?