# AI API outages on August 4, 2026 (UTC)

Measured every 5 minutes from 4 regions. An outage is attributed to a provider only when it failed from 2+ regions at once and no other provider failed in the same minutes.

## Was the friendli API down on August 4, 2026?

Yes, measured. 12 outage(s) attributed to friendli: host probes failed from up to 4 of 4 regions at once while no other provider was failing, first at 02:00 UTC, at least 89 minutes observed in total. At least once friendli's own host answered with an HTTP 5xx.

| UTC | What failed | Regions | Duration (lower bound) | Failed probes | Verdict |
| --- | --- | --- | --- | --- | --- |
| 00:38-01:05 | host probes | ap-tokyo eu-hetzner sa-east | >=27 min | 5 | unattributed |
| 02:00-02:20 | host probes | ap-tokyo eu-hetzner us-central | >=20 min | 4 | provider_outage |
| 04:10-04:15 | host probes | sa-east us-central | >=5 min | 3 | provider_outage |
| 04:55-05:00 | host probes | ap-tokyo sa-east | >=5 min | 2 | provider_outage |
| 06:20-06:23 | host probes | ap-tokyo eu-hetzner | >=3 min | 2 | unattributed |
| 06:58-07:25 | host probes | ap-tokyo eu-hetzner sa-east | >=27 min | 4 | unattributed |
| 08:35-10:00 | host probes | ap-tokyo eu-hetzner sa-east us-central | >=85 min | 12 | unattributed |
| 10:48-10:55 | host probes | eu-hetzner us-central | >=7 min | 2 | provider_outage |
| 12:05-12:08 | host probes | ap-tokyo eu-hetzner sa-east us-central | >=3 min | 4 | unattributed |
| 14:10-14:45 | host probes | ap-tokyo eu-hetzner sa-east us-central | >=35 min | 7 | unattributed |
| 16:15-16:35 | host probes | ap-tokyo sa-east us-central | >=20 min | 3 | unattributed |
| 17:13-17:23 | host probes | eu-hetzner us-central | >=10 min | 4 | provider_outage |
| 17:40-18:00 | host probes | ap-tokyo sa-east us-central | >=20 min | 5 | unattributed |
| 18:20-18:27 | host probes | ap-tokyo eu-hetzner us-central | >=7 min | 3 | provider_outage |
| 19:05-19:13 | host probes | ap-tokyo eu-hetzner sa-east us-central | >=9 min | 4 | provider_outage |
| 19:35-19:40 | host probes | ap-tokyo sa-east | >=5 min | 2 | unattributed |
| 19:55-20:00 | host probes | ap-tokyo us-central | >=5 min | 2 | provider_outage |
| 20:15-20:15 | host probes | ap-tokyo eu-hetzner | <5 min | 2 | provider_outage |
| 20:40-20:46 | host probes | eu-hetzner us-central | >=6 min | 2 | provider_outage |
| 21:11-21:40 | host probes | ap-tokyo eu-hetzner sa-east | >=29 min | 5 | unattributed |
| 22:20-22:25 | host probes | sa-east us-central | >=5 min | 2 | unattributed |
| 23:00-23:05 | host probes | sa-east us-central | >=5 min | 3 | provider_outage |
| 23:20-23:30 | host probes | ap-tokyo sa-east us-central | >=10 min | 3 | provider_outage |

Live status now: https://llmlatency.dev/provider/friendli

## Was the Gemini API (Google) down on August 4, 2026?

Yes, measured. 2 outage(s) attributed to Google: requests to gemini-flash-lite-latest failed from up to 4 of 4 regions at once while no other provider was failing, first at 06:40 UTC, at least 8 minutes observed in total. At least once Google's own host answered with an HTTP 5xx.

| UTC | What failed | Regions | Duration (lower bound) | Failed probes | Verdict |
| --- | --- | --- | --- | --- | --- |
| 05:50-05:58 | requests to gemini-flash-lite-latest | ap-tokyo eu-hetzner sa-east us-central | >=8 min | 4 | unattributed |
| 06:15-06:23 | requests to gemini-flash-lite-latest | ap-tokyo eu-hetzner sa-east us-central | >=8 min | 4 | unattributed |
| 06:40-06:48 | requests to gemini-flash-lite-latest | ap-tokyo eu-hetzner sa-east us-central | >=8 min | 4 | provider_outage |
| 07:05-07:13 | requests to gemini-flash-lite-latest | ap-tokyo eu-hetzner sa-east us-central | >=8 min | 4 | unattributed |
| 07:30-07:38 | requests to gemini-flash-lite-latest | ap-tokyo eu-hetzner sa-east us-central | >=8 min | 4 | unattributed |
| 07:55-08:03 | requests to gemini-flash-lite-latest | ap-tokyo eu-hetzner sa-east us-central | >=8 min | 4 | unattributed |
| 08:20-08:28 | requests to gemini-flash-lite-latest | ap-tokyo eu-hetzner sa-east us-central | >=8 min | 4 | unattributed |
| 12:05-12:05 | requests to gemini-flash-lite-latest | ap-tokyo us-central | <5 min | 2 | unattributed |
| 13:20-13:28 | requests to gemini-flash-lite-latest | ap-tokyo eu-hetzner sa-east | >=8 min | 3 | unattributed |
| 13:45-13:45 | requests to gemini-flash-lite-latest | ap-tokyo us-central | <5 min | 2 | provider_outage |
| 23:45-23:45 | requests to gemini-flash-lite-latest | ap-tokyo sa-east | <5 min | 2 | unattributed |

Live status now: https://llmlatency.dev/provider/google

## Was the Claude API (Anthropic) down on August 4, 2026?

Not by our measurements. No failure of Anthropic was seen from two or more regions at once. Anthropic's own status page opened 3 entries that day (self-reported, listed below).
- self-reported 06:16 UTC: Elevated errors on Claude Sonnet 5 (minor)
- self-reported 11:27 UTC: Elevated errors on Claude Sonnet 5 (minor)
- self-reported 20:48 UTC: Elevated errors across many models (major)

Live status now: https://llmlatency.dev/provider/anthropic

## Was the yi-01ai API down on August 4, 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.

| UTC | What failed | Regions | Duration (lower bound) | Failed probes | Verdict |
| --- | --- | --- | --- | --- | --- |
| 04:30-04:38 | host probes | ap-tokyo eu-hetzner | >=8 min | 2 | unattributed |
| 07:40-07:45 | host probes | ap-tokyo us-central | >=5 min | 2 | unattributed |

Live status now: https://llmlatency.dev/provider/yi-01ai

Source: https://llmlatency.dev/incidents/2026-08-04 · CC BY 4.0