Gladia

Processing queues are empty and pre-recorded transcription processing times are back to normal. Real-time and live streaming sessions are operating normally. During this incident, some customers experienced elevated error rates, timeouts, and degraded processing times on pre-recorded, real-time, and live streaming endpoints. A detailed post-mortem will be shared in the coming days. We sincerely apologize for the disruption and thank you for your patience.

We are still facing performance issues due to reduced infrastructure capacity, average processing times have degraded. Across all accounts, average turnaround is currently around 28 minutes. Actual times still vary by account depending on your current transcription queue and your concurrent processing limit. All transcriptions are still being processed. Some may simply complete more slowly than usual. Our teams continue to work on restoring full capacity.

We are still facing performance issues due to reduced infrastructure capacity, but average processing times have improved. Across all accounts, average turnaround is currently around 12 minutes. Actual times still vary by account depending on your current transcription queue and your concurrent processing limit. All transcriptions are still being processed. Some may simply complete more slowly than usual. Our teams continue to work on restoring full capacity.

We are still facing performance issues due to reduced infrastructure capacity. Processing times vary by account: they depend on your current transcription queue and your concurrent processing limit. Across all accounts, average turnaround is currently around 22 minutes. All transcriptions are still being processed and are not lost. Some may simply complete more slowly than usual. Our teams continue to work on restoring full capacity.

We are still facing performance issues, requests might take a few minutes to be processed.

Incident is now resolved. The average transcription time is less than 1 minute

Uploads are working normally again. Processing capacity is still reduced while we clear the backlog from earlier today. Expect around **3 minutes** for new jobs during peak periods, and longer (up to about **20 minutes**) if you have a high volume waiting in the queue. We are continuing to improve capacity and will update again as times return to normal.

The network issue is now fully resolved. Uploads and API submissions should work normally again. We are still clearing the backlog that built up during the disruption, and we are working to catch up. New jobs are typically completing in around **5 minutes**. Jobs that were already waiting, or accounts with a large queue, can still see longer processing times until that backlog drains. We will keep monitoring and share another update as processing returns to normal.

We have identified the root cause: an external tool generated a traffic spike and a network incident. This delayed transcription processing and impacted our network, which also explains the upload latency. Service is progressively returning to normal. We continue to monitor and will share a final update once everything is confirmed stable.

We are currently experiencing degraded performance on our Europe API. Transcription processing is slower than usual, and file uploads may also be affected. Our team is actively working on a resolution. We will share another update as soon as more information is available.

We are currently experiencing degraded performance on our Europe API. Transcription processing is slower than usual, and file uploads may also be affected. Our team is actively working on a resolution. We will share another update as soon as more information is available.

This issue has been resolved. Pre-recorded transcription processing times have improved further and are now averaging below 4 minutes and are back to normal.

We have restored our GPU capacity. Due to the earlier incident and ongoing high traffic, pre-recorded transcription processing times remain longer than usual. We continue to monitor the situation closely.

Pre-recorded transcription processing times have come down to around 5 minutes on average and are returning to normal. We are monitoring the situation.

Pre-recorded transcription is taking longer than usual. Average processing time is currently around 6 minutes. Investigation ongoing.

Service is unhealthy

Service has been restored. Pre-recorded transcription was largely unaffected.

Real-time API v2 is experiencing intermittent availability and connection failures.

This incident has been resolved.

Service is unhealthy

Translation service was scaled to accommodate heavy load.

Translation service was overloaded.

Service is unhealthy

This incident has been resolved.

Service is unhealthy

This incident has been resolved.

Service is unhealthy

This incident has been resolved.

Service is unhealthy

This incident has been resolved.

Service is unhealthy

This incident has been resolved.

Service is unhealthy

This incident has been resolved.

Service is unhealthy

The incident has been resolved. All pending jobs have been processed successfully.

We've identified to the root cause. We are in recovery and monitoring the situation.

There is a connectivity issue causing huge slow downs in transcriptions. We are working on identifying the root cause.

This incident has been resolved.

Service is unhealthy

Timeframe: 5:02 PM – 5:08 PM Paris Time (CET/CEST) Update: We experienced a brief period of elevated latency and intermittent timeouts affecting the Gladia API. Our engineering team monitored the situation, and performance has since fully stabilized. All systems are now operating normally. We apologize for any disruption this may have caused.

This incident is now resolved. Pre-recorded transcription in Europe is operating normally. Average end-to-end latency for new jobs in the EU pipeline is approximately **25 seconds**, with queue wait under **10 seconds** and transcription completing in about **20 seconds** on average. These values are in line with our usual operating range. Jobs submitted during the earlier slowdown have been processed. Service is fully restored. We apologize for the disruption. If you continue to experience issues, please contact us at [support@gladia.io](mailto:support@gladia.io).

We have applied mitigations and pre-recorded transcription performance in Europe is now back to normal levels for new jobs. Average end-to-end latency for new transcriptions in the EU pipeline is approximately **3 minutes**, in line with our usual operating range. Queue wait is under **2 minutes**, and transcription itself completes in under **30 seconds** on average. Jobs submitted during the earlier slowdown are still being processed and will complete over the coming hours as the backlog clears. No action is required on your side for those requests. We continue to monitor the pipeline and will post a final update once all backlog has cleared and performance is confirmed stable.

We are still investigating degraded performance on pre-recorded transcription in Europe. On our side, current average end-to-end latency for jobs in the EU pipeline is about **15 minutes**. Most of that time is spent waiting in the processing queue (~10 minutes), with the actual transcription step taking around **2 minutes** on average. We will keep the status page updated as the situation evolves. You can subscribe to alerts there if you have not already.

Service is unhealthy

This incident has been resolved.

Service is unhealthy

This incident is now resolved, all services are back to normal. All components are fully operational: Pre-Recorded, Add-ons. Current processing time for pre-recorded transcriptions is approximately 2min.

Service is unhealthy

Incident completed. All of our services are now operational.

A network disruption in the Europe infrastructure triggered cascading failures that ultimately affected database connectivity and API availability. Service was restored after isolating the affected component and performing a database failover.

Corrective actions are deployed. Impacted services: Real-Time v2, Add-ons, Pre-Recorded, Real-Time v. Expected resolution: around **9:45 UTC**.

Network instability API on Pre-recorded and Real Time.

This incident has been resolved.

Service is unhealthy

This incident has been resolved.

Service is unhealthy

This incident has been resolved.

Service is unhealthy

This incident has been resolved.

Service is unhealthy

This incident has been resolved.

Service is unhealthy

This incident has been resolved.

Service is unhealthy

This incident has been resolved.

Service is unhealthy