Parallel

This incident is now resolved.

We have identified the issue and are monitoring fixes.

This issue is now resolved and all intermittent errors have subsided.

Intermittent 401 errors have been observed on the Search API endpoint.

Tasks from the 11:45 am to 12:45 pm are finishing now. Latencies have dropped back down.

Incident is mitigated for new task api requests. A small number of API requests between 11:45 am to 12:45 pm are stuck in a queued state, which will be resolved in the next hour.

Seeing elevated request times due to a spike in traffic, bringing up more capacity to reduce the effect. Expected to resolve in ~20 min.

Traffic has stabilized.

We experienced a brief surge in unusual traffic that caused elevated 503 error rates and temporarily affected service availability for some customers. The traffic surge has subsided, and service availability has returned to normal.

Customers who added credits may not have seen their balance updated while the incident was ongoing.

Prepaid customers may experience issues adding new funds or auto-refreshing balances.

Incident resolved. Our infra is back up and all monitors are sending events.

Monitors are not executing for some customers. Customers won't be charged for any missed executions. All monitors will be backfilled with past events once we're back up.

We have identified and resolved the issue.

FindAll entity search started to return no results for company entity type.

We have expanded our load balancer resources and added targeted alerts to prevent this from happening again.

We've identified the issue and are expanding our load balancer resources.

Some customers' HTTP requests are receiving a 503 response code.

Monitor webhooks have been restored to normal functionality

Monitor webhooks are not triggering. Monitor runs are still succeeding and the triggered events can be obtained via an api call the /events endpoint.

We experienced a disruption due to a DNS configuration issue affecting search-mcp.parallel.ai. This resulted in intermittent errors in the Search MCP service for approximately 10 minutes. Following restoration, some users encountered OAuth reconnection errors when attempting to use Search MCP for an additional period of up to 30 minutes.

This issue is resolved.

The issue was contained to a small percentage of runs & is in the process of resolution.

The Ultra models are seeing high failure rates in the Task API

The issue has been identified and resolved.

Task API webhook are not firing and some tasks are being marked as failed as a result.

Task wait times are back to normal.

We have fixed the underlying issue and processing the backlog of tasks. Wait times should be back to normal.

Tasks that were enqueued between 1:49PM PT to 3:19PM PT are currently waiting to be scheduled. Tasks enqueued since are being executed but may experience longer than usual wait times.

We rolled back the change, all systems are operational now.

[api.parallel.ai](http://api.parallel.ai "api.parallel.ai") was down from 14:07 PST to 14:17 PST due to a bad deployment. We rolled the change back, and all systems are operational now. Apologies for the inconvenience.

We have now resolved this incident.

We are seeing Chat API responsiveness in-Platform reduced. We are investigating.

We have deployed a fix for this issue.

We have resolved the issue.

Some users are experiencing errors on log in as well as account creation. We have identified the root cause.

A minor bug where some users are seeing an issue creating an account.