Is AssemblyAI Down?
Check live AssemblyAI status — speech-to-text transcription, real-time streaming, LeMUR (LLM on audio), speaker diarization, and API. See recent incidents and set up free email alerts.
AssemblyAI — live status
Updated every 5 minutes. Full history at prismix.dev/service/assemblyai.
Quick check: is AssemblyAI down right now?
- Prismix: prismix.dev/service/assemblyai — live status + 30-day uptime + incidents.
- API call:
curl https://prismix.dev/api/v1/statuses | jq '.services[] | select(.id=="assemblyai")' - Direct test:
curl https://api.assemblyai.com/v2/transcript -H "Authorization: YOUR_KEY" -d '{"audio_url":"..."}'— a 200 response confirms the API is reachable.
Common causes of "AssemblyAI not working"
- Transcription stuck in "queued" or "processing" — AssemblyAI downloads the audio file from your URL; if the URL is behind authentication or not publicly accessible, it fails silently and the job stays queued. Use the /v2/transcript/{id} polling endpoint to track status; processing typically takes 15–30% of the audio's duration.
- HTTP 400 "Audio file could not be downloaded" — the audio URL must be publicly accessible with no auth headers required. If your file is private, use AssemblyAI's /v2/upload endpoint to upload the file directly first, then pass the returned upload_url as the audio_url in your transcription request.
- LeMUR responses returning generic answers — LeMUR uses the completed transcript as its context window; if you call LeMUR before transcription finishes, it has no content to work with. Always poll until status=completed before triggering a LeMUR request. Very long transcripts are chunked automatically, but complex queries across multiple chunks may produce less precise answers.
- Speaker diarization showing wrong speaker count — set the speakers_expected parameter to the known number of speakers; without it, AssemblyAI auto-detects speaker count which is less accurate with overlapping voices. Note that diarization requires a minimum of approximately 1 minute of audio to function reliably.
- Webhook not firing on completion — AssemblyAI POSTs to your webhook_url when transcription completes. Verify your endpoint is publicly accessible (not localhost), responds with HTTP 200 within 5 seconds, and that you set the webhook_url field on the original transcript creation request. Timeouts on your end will cause AssemblyAI to stop retrying.
- Rate limit 429 errors — the free tier allows 5 concurrent transcriptions; paid tiers scale based on plan. Implement exponential backoff when you receive a 429 and check your current usage and concurrency limits at app.assemblyai.com.
Set up free email alerts for AssemblyAI
- 1
Sign in to Prismix
Go to prismix.dev/sign-in — email OTP or GitHub sign-in.
- 2
Star AssemblyAI
On prismix.dev/service/assemblyai, click the ☆ star icon.
- 3
Alerts are live
You'll get an email within minutes of any status change.
Get an email the next time AssemblyAI goes down
Outage alerts for AssemblyAI, straight to your inbox. No account needed, unsubscribe in every email.
Watching more than one service? A free account covers 5 services + a daily digest — and Pro is currently free.
Monitor other AI audio & speech tools?
Full status dashboard: prismix.dev/status