Workspace Agents API trigger consistently returns HTTP 409 in a ChatGPT Business workspace

ChatGPT
,
1

We are trying to use the Workspace Agents API trigger in a ChatGPT Business workspace, but every trigger request consistently returns:

{ "error": { "message": "The workspace agent trigger is not currently available.", "type": "invalid_request_error", "param": null, "code": null } }

HTTP status: 409 Conflict

What is confusing is that the API requests appear in the agent’s Activity tab, but each run is marked:

Could not complete task

This suggests that the request reaches the workspace agent system, but the run cannot be started or completed.

Setup

  • ChatGPT Business workspace

  • Workspace administrator account

  • Agent is live and up to date

  • API channel is enabled

  • Access token status is Active

  • Access token was created with the Workspace Agents scope

  • The agent works normally through the ChatGPT web interface

  • Tools, apps, and skills also work when the agent is used through the web interface

    Minimal request

    curl -i -sS -X POST \ "``https://api.chatgpt.com/v1/workspace_agents/``<TRIGGER_ID>/trigger" \ -H "Authorization: Bearer <WORKSPACE_AGENT_ACCESS_TOKEN>" \ -H "Content-Type: application/json" \ --data '{"input":"Hello world"}'

We also tested the request with:

-H "OpenAI-Beta: workspace_agent_runs=v1" -H "Idempotency-Key: <UNIQUE_VALUE>"

The result remained exactly the same: HTTP 409.

Troubleshooting already completed

  1. Confirmed that the access token is populated and valid.

  2. Created a new access token with the Workspace Agents scope.

  3. Confirmed that the token is shown as Active.

  4. Removed and recreated the API channel.

  5. Tested with and without the beta and idempotency headers.

  6. Tested a minimal {"input":"Hello world"} request.

  7. Tested the original agent through the ChatGPT web interface successfully.

  8. Created a completely new minimal agent specifically for API testing.

  9. Added a new API channel to the new agent.

  10. Triggered the new agent using its newly generated trigger ID.

  11. Received the same HTTP 409 response from the completely new agent.

  12. Confirmed that the failed API-triggered runs appear in the agent Activity tab as Could not complete task.

Workspace settings

We cannot find any option in the ChatGPT Business workspace or admin settings to enable, disable, pause, or provision Workspace Agent API triggers.

The UI allows us to:

  • Create Workspace Agent access tokens

  • Select the Workspace Agents token scope

  • Add and enable an API channel

  • Obtain an agtch_... trigger ID and API URL

However, there is no separate workspace-level API trigger setting available to us.

The access-token details also show:

  • Use with: Workspace Agents

  • Status: Active

  • Last used: Not available

The “Last used” value remains unavailable even after multiple authenticated requests. The requests return HTTP 409 rather than an authentication error, so authentication appears to be accepted before the trigger becomes unavailable.

OpenAI Support indicated that the expected response should be 202 Accepted, and that a persistent HTTP 409 likely indicates either:

  • Workspace-level availability or feature gating

  • A trigger provisioning problem

Because both the original agent and a completely new test agent return the same error, this appears more likely to be a workspace-level issue rather than a problem with one trigger.

Any confirmation from another ChatGPT Business workspace owner would be very helpful.