Skip to main content
Deepgram logo

Is Deepgram Down?

No — Deepgram is up

Reachable from all 7 checked regions

Average response time: 218ms

Last checked · checks run every 6 hours

Official status page: https://status.deepgram.com

Deepgram uptime

100%
Last 7 days
100%
Last 30 days
100%
Last 90 days
236ms
Avg response, 30 days

Measured from multiple regions every 6 hours. Percentages count only checks that returned an availability answer — 8 days measured so far. A dash means that window does not yet hold enough measured days to publish a figure.

30-day history

8-day clean streak
Sep 14: no data
Sep 15: no data
Sep 16: no data
Sep 17: no data
Sep 18: no data
Sep 19: no data
Sep 20: no data
Sep 21: no data
Sep 22: no data
Sep 23: no data
Sep 24: no data
Sep 25: no data
Sep 26: no data
Sep 27: no data
Sep 28: no data
Sep 29: no data
Sep 30: no data
Oct 1: no data
Oct 2: no data
Oct 3: no data
Oct 4: no data
Oct 5: no data
Oct 6: 100.00% uptime, 24 checks
Oct 7: 100.00% uptime, 32 checks
Oct 8: 100.00% uptime, 32 checks
Oct 9: 100.00% uptime, 32 checks
Oct 10: 100.00% uptime, 32 checks
Oct 11: 100.00% uptime, 32 checks
Oct 12: 100.00% uptime, 32 checks
Oct 13: 100.00% uptime, 23 checks
Sep 14 Today
No downtime Partial Downtime Not measurable No data

Reachability by region

Each region runs its own request from a different part of the world. A service can be up for one continent and down for another, which is usually the first sign of a routing or CDN problem.

arn
134ms
DNS 51ms TCP 1ms TLS 11ms TTFB 126ms
dfw
240ms
DNS 72ms TCP 0ms TLS 53ms TTFB 191ms
ewr
182ms
DNS 67ms TCP 1ms TLS 23ms TTFB 165ms
fra
188ms
DNS 15ms TCP 1ms TLS 10ms TTFB 180ms
gru
67ms
DNS 21ms TCP 1ms TLS 8ms TTFB 60ms
iad
242ms
DNS 47ms TCP 2ms TLS 8ms TTFB 105ms
jnb
478ms
DNS 231ms TCP 0ms TLS 78ms TTFB 405ms

What Deepgram does

Deepgram is a voice AI API. Developers send recorded files to its batch speech-to-text endpoint, stream live audio over WebSockets for real-time transcripts, generate speech with its Aura and Flux text-to-speech models, and build conversational bots with the Voice Agent API. Requests go to api.deepgram.com by default. EU customers can point the same API keys and SDKs at api.eu.deepgram.com to keep processing inside the EU.

What an outage looks like

REST calls to endpoints such as /v1/listen return 500, 502, 503 or 504 errors, and a 503 body reads Please try again later. Live streams fail to open or close with WebSocket code 1011 and a NET-0000 error, meaning Deepgram sent no transcript within its timeout window. Incidents usually name one product: Flux TTS errors, slow Batch STT, failing Voice Agent LLM providers or 5XX errors from v1/auth/grant.

What to do about it

Check status.deepgram.com, which has separate rows for the Batch, Streaming, TTS and Voice Agent APIs, the Developer Console and self-hosted license validation. Retry 429 responses with exponential backoff; 5xx errors may also succeed on retry. For live streams, buffer audio while you open a new WebSocket, then offset the new transcript timestamps, which restart at zero. If a problem outlasts the incident, email support@deepgram.com with the dg-request-id header.

Is it down for everyone, or just you?

If this page says Deepgram is up but it is not loading for you, the problem is between you and them. Run a check against any URL from up to 17 cities across three regions to find out where it breaks.

Test it yourself

Related services

ElevenLabs
ElevenLabs
Twilio
Twilio
C
Claude
AWS
AWS

Deepgram outage FAQ

Is Deepgram down for everyone or just me?
Check status.deepgram.com first; Deepgram's own support page says it is the quickest way to rule out a problem on their side, and that you do not need to open a ticket to report an incident. If it is green, the dg-error header on a failed WebSocket upgrade explains why the connection was refused. The checks on this page probe deepgram.com, the company website, so an API fault on api.deepgram.com can happen while they stay green.
Will Deepgram retry my callback if my server is down?
Yes. When you submit audio with a callback URL, Deepgram sends the result as a POST to that URL, and if your server answers with anything outside 200-299 it retries up to 10 times, 30 seconds apart. That gives you about five minutes to recover before the result is lost. Make sure the callback server accepts large bodies: a body-size limit that answers 413 Request Entity Too Large is a common reason a transcript never arrives.
Does the Deepgram EU endpoint fail over to the US during an outage?
No. Requests to api.eu.deepgram.com are never routed outside the EU, so if the EU region is unavailable they fail rather than fall back to another region. Switching to the global api.deepgram.com endpoint during an EU incident would move processing out of the EU, so check your data-residency obligations first. Incidents that affect only the EU are marked Deepgram EU API in their titles on the status page.
Does a Deepgram outage affect self-hosted deployments?
Only through licensing. Self-hosted Deepgram contacts the Deepgram license server to validate components and report usage, and the status page has a License Validation row for it. With the License Proxy, running services keep working for a limited time on cached trust if the license server is unreachable, but a new proxy instance cannot start without it. So avoid replacing or adding License Proxy instances until the licensing incident clears.
How long do Deepgram outages usually last?
Most are resolved within a few hours. Of the 50 most recent incidents on Deepgram's status page, the median was resolved in a little over two hours, and 31 were closed within four hours, though nine stayed open for more than a day. You can subscribe on status.deepgram.com to updates by email, text message, Slack, webhook or RSS instead of refreshing the page.

How we measure this

  • We request Deepgram's public endpoint every 6 hours from Fly.io regions across six continents — the most recent check ran from 7 of them.
  • A region counts as down only when it gets no usable HTTP response. A 403 or 429 means the origin answered and refused us, which we report as blocked, never as an outage.
  • A single failing region is treated as probe noise. We only change the verdict when two consecutive cycles agree.
  • Response times average only the regions that actually served the page, so a timeout never inflates the number.
  • Where Deepgram publishes an official status feed we read it too. An all-clear from the vendor can soften an unconfirmed degradation; a vendor-declared outage only worsens our verdict when our own checks corroborate it.

Get alerted when Deepgram goes down

This page refreshes every 6 hours. Your own monitors run as often as every 30 seconds, from up to 17 cities across three regions, and tell you the moment something breaks.

Start Free Monitoring
Free plan available No credit card required