DigitalOcean incident: Response Degradation Impacting Kimi-K3

Affected components

• Model Services

1. resolvedJul 30, 2026, 06:15 UTC

   Our Engineering team has resolved the issue causing empty or broken responses from Kimi-K3. If you continue to experience any problems, please open a ticket with our Support team. We apologize for any inconvenience this may have caused.

2. investigatingJul 29, 2026, 21:41 UTC

   Our Engineering team is investigating reports of empty or broken responses impacting Kimi-K3. At this point, users may experience degraded response quality, specifically empty or broken outputs (including repeated characters), when querying the Kimi-K3 service. We apologize for the inconvenience and will share an update once we have more information.

3. investigatingJul 29, 2026, 17:56 UTC

   We are currently investigating this issue.

What APIStatus.watch saw

Our 5-minute checks didn't record a change in DigitalOcean's overall status around this incident. Smaller or regional incidents often leave a provider's overall status green.

Other services with incidents at the same time

Overlapping incidents aren't necessarily related.

• Anthropic: Elevated errors across all modelsCritical · Jul 29, 2026, 19:49 UTC
• Snowflake: INC20000115Critical · Jul 30, 2026, 03:00 UTC
• HashiCorp Vault: HCP Terraform Workspace Runs Failing with IAM Errors After AWS Provider UpgradeMajor · Jul 29, 2026, 12:36 UTC
• Resend: Elevated error rates on email sending, API, and dashboardMajor · Jul 29, 2026, 21:07 UTC
• Anthropic: Elevated errors across many modelsMajor · Jul 30, 2026, 05:57 UTC
• RevenueCat: Newly created apps error with "The key is not valid or is not compatible with the Bundle ID of your app"Minor · Jul 27, 2026, 10:14 UTC
• Grafana Cloud: Grafana Cloud non-billing usage metrics gaps in select regionsMinor · Jul 29, 2026, 18:49 UTC
• MongoDB Atlas: Delayed cluster configuration affecting free and flex usersMinor · Jul 29, 2026, 18:55 UTC
• GitHub: Incident with Copilot AI Model ProvidersMinor · Jul 29, 2026, 20:07 UTC
• Grafana Cloud: IRM Performance Degradation in EU RegionMinor · Jul 29, 2026, 21:46 UTC

How often does DigitalOcean have incidents?

DigitalOcean reported 20 incidents in the last 90 days, 3 of them major or critical. A typical incident lasted 5h 35m. See DigitalOcean's uptime and incident history