API Reliability Report: Q3 2026

Key findings

• Shippo and OVHcloud tied for the most major or critical incidents, with 43 each.
• The longest major incident was Elasticsearch's “Elasticsearch 9.5.0 contains a query-correctness defect” on Aug 5, 2026, which lasted 6d 10h.
• Half of all incidents were resolved within 2h 42m.
• Cloud had the most major incidents of any category (71 across 7 services).
• 12 of 71 services published no major incidents at all.

Major and critical incidents by service

Top 15, as declared by each provider. Hover or focus a bar for all incidents and hours.

1. Shippo
   43

   43 major or critical

   Shippo · 67 incidents in total · 156h under major incidents

2. OVHcloud
   43

   43 major or critical

   OVHcloud · 54 incidents in total · 475.7h under major incidents

3. Anthropic
   32

   32 major or critical

   Anthropic · 87 incidents in total · 148.7h under major incidents

4. GitHub
   26

   26 major or critical

   GitHub · 65 incidents in total · 79h under major incidents

5. Grafana Cloud
   24

   24 major or critical

   Grafana Cloud · 65 incidents in total · 96.4h under major incidents

6. Cursor
   23

   23 major or critical

   Cursor · 87 incidents in total · 51.5h under major incidents

7. Elasticsearch
   21

   21 major or critical

   Elasticsearch · 26 incidents in total · 432.1h under major incidents

8. Supabase
   18

   18 major or critical

   Supabase · 51 incidents in total · 182.9h under major incidents

9. Snowflake
   18

   18 major or critical

   Snowflake · 23 incidents in total · 42.1h under major incidents

10. Discord
    16

    16 major or critical

    Discord · 17 incidents in total · 21.9h under major incidents

11. Sentry
    15

    15 major or critical

    Sentry · 25 incidents in total · 45.2h under major incidents

12. Vercel
    10

    10 major or critical

    Vercel · 26 incidents in total · 9.4h under major incidents

13. Render
    10

    10 major or critical

    Render · 17 incidents in total · 18.3h under major incidents

14. HashiCorp Vault
    8

    8 major or critical

    HashiCorp Vault · 15 incidents in total · 25.4h under major incidents

15. Inngest
    8

    8 major or critical

    Inngest · 14 incidents in total · 99.1h under major incidents

Longest major incidents

As rated by each provider. Not every major incident is a full outage: some are degradations or advisories (follow the link for the provider's own timeline).

1. 1Elasticsearch: Elasticsearch 9.5.0 contains a query-correctness defectAug 5, 20266d 10h
2. 2Elasticsearch: Elasticsearch 9.5.1: false-positive matches in certain boolean queriesAug 14, 20266d 7h
3. 3Anthropic: Degraded functionality for Claude Cowork on WindowsSep 10, 20264d 3h
4. 4Elasticsearch: Increased error rates for microsoft-multilingual-e5-largeAug 13, 20264d 3h
5. 5Supabase: Increased response times for requestsAug 27, 20263d 22h
6. 6OVHcloud: [GRA][Network] - Public network incident notificationJul 9, 20263d 22h
7. 7OVHcloud: [US-EAST-VA][Storage] - Object Storage incident notificationSep 25, 20263d 21h
8. 8Inngest: Issues Affecting Function Triggering from EventsJul 10, 20263d 2h
9. 9Elasticsearch: Elastic Cloud - Email MFA Delivery Delays (Gmail)Jul 21, 20262d 7h
10. 10Shippo: Axlehire tracking disruptionsSep 20, 20261d 19h