HomeIncidents
Data API

AI API outages on September 30, 2026

UTC day · measured every 5 minutes from 4 regions · method

On September 30, 2026 (UTC) our probes attributed 5 outage(s) to 4 provider(s): friendli API: 2; Gemini API (Google): 1; SambaNova API: 1; sarvam API: 1. Providers’ own status pages opened 3 entries (OpenAI).

This page is history. Is it down right now?

Live status, updated every 30 minutes: friendli API, Gemini API (Google), SambaNova API, sarvam API, OpenAI API · all AI APIs, last 72 hours

Was the friendli API down on September 30, 2026?

Yes, measured. 2 outage(s) attributed to friendli: host probes failed from up to 2 of 4 regions at once while no other provider was failing, first at 11:49 UTC, at least 2 minutes observed in total.

UTCWhat failedRegionsDurationFailed probesVerdictError
11:49–11:50host probes2 of 4: Asia (Tokyo), South America (São Paulo)≥1 min2provider outagetimeout
12:19–12:20host probes2 of 4: South America (São Paulo), US (Central)≥1 min2provider outagetimeout

Current friendli API status

Was the Gemini API (Google) down on September 30, 2026?

Yes, measured. 1 outage(s) attributed to Google: requests to gemini-flash-lite-latest failed from up to 2 of 4 regions at once while no other provider was failing, first at 07:55 UTC, at least 8 minutes observed in total.

UTCWhat failedRegionsDurationFailed probesVerdictError
06:05–06:05requests to gemini-flash-lite-latest3 of 4: Asia (Tokyo), South America (São Paulo), US (Central)<5 min3shared with 1 other provider(s) — not attributedtimeout
07:55–08:04requests to gemini-flash-lite-latest2 of 4: Europe (Germany), South America (São Paulo)≥8 min2provider outagetimeout

Current Gemini API (Google) status

Was the SambaNova API down on September 30, 2026?

Yes, measured. 1 outage(s) attributed to SambaNova: host probes failed from up to 2 of 4 regions at once while no other provider was failing, first at 07:00 UTC, each shorter than one 5-minute probe cycle. SambaNova’s status page lists nothing that day.

UTCWhat failedRegionsDurationFailed probesVerdictError
07:00–07:00host probes2 of 4: Europe (Germany), US (Central)<5 min2provider outagetimeout

Current SambaNova API status

Was the sarvam API down on September 30, 2026?

Yes, measured. 1 outage(s) attributed to sarvam: host probes failed from up to 3 of 4 regions at once while no other provider was failing, first at 20:07 UTC, at least 3 minutes observed in total. At least once sarvam’s own host answered with an HTTP 5xx.

UTCWhat failedRegionsDurationFailed probesVerdictError
20:07–20:10host probes3 of 4: Europe (Germany), South America (São Paulo), US (Central)≥3 min3provider outagehttp 502; timeout

Current sarvam API status

Was the OpenAI API down on September 30, 2026?

Not by our measurements. No failure of OpenAI was seen from two or more regions at once; 1 single-region failure(s) are not called an outage. OpenAI’s own status page opened 3 entries that day (self-reported, listed below).

Self-reported on OpenAI’s status page, not verified by our probes:

  • minor Elevated errors in ChatGPT Space Pages · resolved Oct 1, 23:44 UTC
  • minor Elevated error rates for ChatGPT Pro and Plus users · resolved Sep 30, 13:23 UTC
  • minor Elevated latency for some API requests · resolved Sep 30, 20:35 UTC

Current OpenAI API status

29 single-region failure(s) were also recorded that day (google 9, doubao 5, iflytek 5, sensenova 4, stepfun 3, sarvam 1, sambanova 1, openai 1). A failure seen from one region cannot be separated from a problem on that network path, so none of them is called an outage.

How is an outage told apart from a network problem?

Each provider is probed every five minutes from 4 independent regions. A failure is attributed to the provider only when it is seen from two or more regions at once and no other provider fails in the same minutes; a failure shared with other providers has a cause above all of them and is not blamed on anyone. Durations are lower bounds: we see the first and last failed probe, not the real start and end. Status-page entries are the provider’s own words, shown next to the measurement, not instead of it.

· Last 72 hours ·

Machine-readable: incidents.json · Licence: CC BY 4.0