Degraded performance — Flux streaming Speech-to-Text

What to do

• Check whether you use Deepgram Streaming API. Check your stack

  Paste a package.json, lockfile, Terraform or a list of domains at down.guide/stack to see every provider you depend on.

• Confirm your retries, queues and scheduled jobs caught up after the 1d 3h it lasted.

  Anything that failed between 2026-06-15 11:00 UTC and 2026-06-16 14:00 UTC may need a replay.

• Subscribe to Deepgram's feed to hear about the next one first. RSS feed

  https://down.guide/feed?provider=deepgram works in Slack (/feed subscribe), Discord feed bots and any RSS reader.

Updates from Deepgram 3

1. investigatingJun 15, 2026, 11:00 AM UTC

   Some streaming Speech-to-Text connections on the Flux model established successfully but then terminated a few seconds into the session due to an internal error, before returning a final transcript. Applications relying on final transcripts (e.g. IVR/voice-agent flows) would have seen sessions open with no transcript returned.

   provider

2. identifiedJun 16, 2026, 1:00 PM UTC

   We identified a degradation affecting a subset of Flux streaming Speech-to-Text sessions, where connections terminated before returning a transcript. We've isolated it to specific degraded hosts and are mitigating.

   provider

3. resolvedJun 16, 2026, 2:00 PM UTC

   This incident has been resolved. A degradation affecting a subset of streaming Speech-to-Text (Flux) sessions, where connections terminated before returning a transcript, has been mitigated and service has been operating normally since. We're continuing to monitor and apologize for the disruption.

   provider

Corrections

None so far.

Something wrong on this page? Report an error or send us a note. Every fix is listed here.