Overview
- OpenAI, Anthropic and xAI all reported elevated errors and partial service failures Thursday that interrupted chats, logins, developer APIs, file uploads, image generation and other core features.
- Engineers applied mitigations and brought systems back online in stages, with OpenAI reporting a routing fix during the morning and Anthropic and xAI resolving most problems within minutes to a few hours.
- No single external cause has been confirmed by the companies, although xAI/SpaceXAI said an outage at its Memphis compute center affected Grok and several compute partners.
- Outage trackers logged large spikes in user reports — tens of thousands for ChatGPT at peak and thousands for other services — causing missed deadlines and disrupted work for students, developers and businesses.
- The episode adds to a series of 2026 service degradations and sharpens questions about uptime guarantees, enterprise SLAs and the practical costs of building true multi‑provider redundancy.