Significant Slowdown in Whisper and GPT-4o Transcribe APIs (since May 2026)

API Bugs
, ,
1

Hi everyone,

I’ve been using the OpenAI transcription models (whisper-1, gpt-4o-transcribe, and gpt-4o-mini-transcribe) in a production environment for about a year without any major issues.

However, since around the beginning of this week (approx. May 4–6, 2026), I’ve noticed a significant and consistent slowdown across all transcription models. Processing times have roughly doubled compared to what I was experiencing before.

This issue affects:

  • whisper-1

  • gpt-4o-transcribe

  • gpt-4o-mini-transcribe

The slowdown is not limited to a specific file or request. It appears to be consistent across different audio inputs and persists over multiple days.

Important notes:

  • No changes were made to my code or infrastructure

  • The same audio files that were previously processed quickly are now much slower

  • The issue occurs consistently, not just intermittently

This is currently impacting a production workflow, so the increased latency is becoming a serious problem.

I wanted to ask:

  • Is anyone else experiencing similar performance issues?

  • Are there any known infrastructure changes, regional slowdowns, or ongoing incidents affecting transcription services?

  • Has anything changed recently regarding model prioritization or performance?

Any insights would be greatly appreciated.

Thanks in advance.

2

Problem solved! I had connection errors (low uploadspeed) so I had to change some settings on my router. Thanks for your help!