AI API outages on August 19, 2026
UTC day · measured every 5 minutes from 4 regions · method
On August 19, 2026 (UTC) our probes attributed 1 outage(s) to 1 provider(s): deepinfra API: 1. Providers’ own status pages opened 3 entries (Anthropic, baseten, nebius).
This page is history. Is it down right now?
Live status, updated every 30 minutes: deepinfra API, Claude API (Anthropic), Gemini API (Google), baseten API, nebius API, 01.AI API · all AI APIs, last 72 hours
Was the deepinfra API down on August 19, 2026?
Yes, measured. 1 outage(s) attributed to deepinfra: host probes failed from up to 4 of 4 regions at once while no other provider was failing, first at 03:07 UTC, at least 10 minutes observed in total. At least once deepinfra’s own host answered with an HTTP 5xx.
| UTC | What failed | Regions | Duration | Failed probes | Verdict | Error |
|---|---|---|---|---|---|---|
| 03:07–03:17 | host probes | 4 of 4: Asia (Tokyo), Europe (Germany), South America (São Paulo), US (Central) | ≥10 min | 9 | provider outage | http 502 |
Was the Claude API (Anthropic) down on August 19, 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:
- minor Degraded performance for Claude Opus 5 and Claude Haiku 4.5 · resolved Aug 19, 11:02 UTC
Current Claude API (Anthropic) status
Was the Gemini API (Google) down on August 19, 2026?
Not attributably. 3 multi-region failure(s) of Google coincided with failures of other providers, so the cause sat above Google and is not blamed on it.
| UTC | What failed | Regions | Duration | Failed probes | Verdict | Error |
|---|---|---|---|---|---|---|
| 10:02–10:10 | requests to gemini-flash-lite-latest | 2 of 4: Europe (Germany), US (Central) | ≥7 min | 2 | shared with 1 other provider(s) — not attributed | The read operation timed out |
| 13:48–13:55 | requests to gemini-flash-lite-latest | 2 of 4: Europe (Germany), US (Central) | ≥7 min | 2 | shared with 1 other provider(s) — not attributed | The read operation timed out |
| 15:02–15:10 | requests to gemini-flash-lite-latest | 2 of 4: Europe (Germany), US (Central) | ≥7 min | 2 | shared with 1 other provider(s) — not attributed | The read operation timed out |
Current Gemini API (Google) status
Was the baseten API down on August 19, 2026?
Not by our measurements. No failure of baseten was seen from two or more regions at once. baseten’s own status page opened 1 entry that day (self-reported, listed below).
Self-reported on baseten’s status page, not verified by our probes:
- minor Elevated errors for moonshotai/Kimi-K3 for US-pinned traffic on Model APIs. **Dedicated In · resolved Aug 19, 02:00 UTC
Was the nebius API down on August 19, 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:
- major Region us-central1 partial degradation · resolved Aug 20, 08:16 UTC
Was the 01.AI API down on August 19, 2026?
Not attributably. 1 multi-region failure(s) of 01.AI coincided with failures of other providers, so the cause sat above 01.AI and is not blamed on it.
| UTC | What failed | Regions | Duration | Failed probes | Verdict | Error |
|---|---|---|---|---|---|---|
| 13:15–14:10 | host probes | 2 of 4: Asia (Tokyo), US (Central) | ≥55 min | 8 | shared with 1 other provider(s) — not attributed | timeout |
51 single-region failure(s) were also recorded that day (yi-01ai 29, google 5, sensenova 5, iflytek 3, sambanova 3, doubao 1, writer 1, stepfun 1, nebius 1, mistral 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.
Previous: Aug 18 · Last 72 hours · Next: Aug 20
Machine-readable: incidents.json · Licence: CC BY 4.0