AI API outages on August 12, 2026

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

On August 12, 2026 (UTC) our probes attributed 6 outage(s) to 1 provider(s): OpenAI API: 6. Providers’ own status pages opened 2 entries (Anthropic, nebius).

This page is history. Is it down right now?

Live status, updated every 30 minutes: OpenAI API, Claude API (Anthropic), nebius API, iflytek API, yi-01ai API · all AI APIs, last 72 hours

Was the OpenAI API down on August 12, 2026?

Yes, measured. 6 outage(s) attributed to OpenAI: host probes failed from up to 4 of 4 regions at once while no other provider was failing, first at 00:40 UTC, at least 166 minutes observed in total. OpenAI’s status page lists nothing that day.

UTCWhat failedRegionsDurationFailed probesVerdictError
00:00–00:05host probes2 of 4: Asia (Tokyo), Europe (Germany)≥4 min2shared with 1 other provider(s) — not attributedtimeout
00:40–00:45host probes2 of 4: South America (São Paulo), US (Central)≥5 min2provider outagetimeout
01:20–02:15host probes4 of 4: Asia (Tokyo), Europe (Germany), South America (São Paulo), US (Central)≥55 min10shared with 1 other provider(s) — not attributedtimeout
03:05–04:20host probes4 of 4: Asia (Tokyo), Europe (Germany), South America (São Paulo), US (Central)≥75 min12shared with 2 other provider(s) — not attributedtimeout
04:50–05:15host probes3 of 4: Asia (Tokyo), Europe (Germany), South America (São Paulo)≥25 min5shared with 1 other provider(s) — not attributedtimeout
05:35–06:00host probes4 of 4: Asia (Tokyo), Europe (Germany), South America (São Paulo), US (Central)≥25 min6shared with 1 other provider(s) — not attributedtimeout
06:25–06:30host probes2 of 4: Europe (Germany), South America (São Paulo)≥4 min2shared with 1 other provider(s) — not attributedtimeout
07:05–07:35host probes4 of 4: Asia (Tokyo), Europe (Germany), South America (São Paulo), US (Central)≥30 min8shared with 1 other provider(s) — not attributedtimeout
08:25–08:40host probes2 of 4: Asia (Tokyo), Europe (Germany)≥16 min4shared with 1 other provider(s) — not attributedtimeout
09:15–09:35host probes3 of 4: Asia (Tokyo), Europe (Germany), US (Central)≥20 min4shared with 1 other provider(s) — not attributedtimeout
09:55–11:20host probes4 of 4: Asia (Tokyo), Europe (Germany), South America (São Paulo), US (Central)≥85 min19shared with 5 other provider(s) — not attributedtimeout
11:40–12:05host probes3 of 4: Asia (Tokyo), South America (São Paulo), US (Central)≥25 min4shared with 2 other provider(s) — not attributedtimeout
12:45–12:55host probes2 of 4: Europe (Germany), South America (São Paulo)≥10 min2shared with 2 other provider(s) — not attributedtimeout
13:40–14:45host probes3 of 4: Asia (Tokyo), Europe (Germany), US (Central)≥65 min8shared with 4 other provider(s) — not attributedtimeout
15:20–15:45host probes3 of 4: Asia (Tokyo), Europe (Germany), South America (São Paulo)≥26 min4provider outagetimeout
16:40–16:50host probes2 of 4: Europe (Germany), US (Central)≥10 min2provider outagetimeout
17:10–17:35host probes3 of 4: Europe (Germany), South America (São Paulo), US (Central)≥25 min6provider outagetimeout
18:00–19:30host probes4 of 4: Asia (Tokyo), Europe (Germany), South America (São Paulo), US (Central)≥90 min15shared with 2 other provider(s) — not attributedhttp 520; timeout
19:55–20:10host probes3 of 4: Europe (Germany), South America (São Paulo), US (Central)≥14 min3provider outagetimeout
20:25–21:50host probes4 of 4: Asia (Tokyo), Europe (Germany), South America (São Paulo), US (Central)≥85 min13provider outagetimeout
22:10–23:45host probes4 of 4: Asia (Tokyo), Europe (Germany), South America (São Paulo), US (Central)≥95 min17shared with 1 other provider(s) — not attributedtimeout

Is the OpenAI API down right now? Live status →

Was the Claude API (Anthropic) down on August 12, 2026?

Not by our measurements. No failure of Anthropic was seen from two or more regions at once. Anthropic’s own status page opened 1 entry that day (self-reported, listed below).

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

Is the Claude API (Anthropic) down right now? Live status →

Was the nebius API down on August 12, 2026?

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

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

Is the nebius API down right now? Live status →

Was the iflytek API down on August 12, 2026?

Not attributably. 1 multi-region failure(s) of iflytek coincided with failures of other providers, so the cause sat above iflytek and is not blamed on it.

UTCWhat failedRegionsDurationFailed probesVerdictError
12:35–13:10host probes2 of 4: Asia (Tokyo), US (Central)≥35 min4shared with 4 other provider(s) — not attributedtimeout

Is the iflytek API down right now? Live status →

Was the yi-01ai API down on August 12, 2026?

Not attributably. 2 multi-region failure(s) of yi-01ai coincided with failures of other providers, so the cause sat above yi-01ai and is not blamed on it.

UTCWhat failedRegionsDurationFailed probesVerdictError
07:05–07:40host probes2 of 4: Asia (Tokyo), Europe (Germany)≥35 min5shared with 1 other provider(s) — not attributedtimeout
11:50–12:10host probes3 of 4: Asia (Tokyo), South America (São Paulo), US (Central)≥20 min5shared with 3 other provider(s) — not attributedtimeout

Is the yi-01ai API down right now? Live status →

57 single-region failure(s) were also recorded that day (yi-01ai 22, openai 7, doubao 7, iflytek 6, friendli 5, google 3, sarvam 2, glm 1, nebius 1, baichuan 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