Notion restores Anthropic models after brief service disruption

Notion briefly cut off access to Anthropic’s models inside Notion AI after degraded performance set off a higher rate of failures for users selecting Opus 4.7 and 4.8. The company said it was disabling use of all Anthropic models in its automated productivity tool, a move that underscored how quickly a single provider problem can ripple through a workplace product embedded in daily collaboration.

The disruption hit a service that sits at the center of Notion’s broader workspace software, where teams rely on AI features to summarize, draft, and organize work. It also landed in the middle of a close partnership between Notion and Anthropic. Anthropic has said Notion served as a design partner in the early development of Claude Code, and that Notion’s engineering teams use Claude Code across the organization.

About twelve hours later, Notion’s head of product, Max Schoening, said he was “astonished” at “the amount of people RT-ing this,” as the post had been reposted around 1,200 times on X. Schoening pushed back on the idea that the outage signaled deeper product quality issues, saying, “The degraded performance was a temporary service disruption. This happens. It happens to Notion, GitHub, AWS, your OpenClaw, and everything in between.” He then said Notion had restored access to Anthropic’s models.

Anthropic said the episode stemmed from infrastructure trouble of its own. In a statement, the company said “a brief infrastructure issue caused elevated errors on multiple Claude models for a short period of time,” and that the problem had since been resolved. Anthropic thanked users for their patience while service was restored.

The incident is a small but revealing snapshot of a larger shift in enterprise software: workplace tools are increasingly built on top of a narrow set of AI model providers, and that concentration creates operational risk whenever one of those providers stumbles. Notion’s rapid disablement of all Anthropic models showed how tightly coupled modern productivity software can be to upstream AI systems, while Anthropic’s response showed how little visibility customers often get beyond a brief explanation after service is already affected.

That matters even more as Anthropic pushes forward with new releases. The company announced Claude Opus 4.8 on May 28, 2026, while positioning itself as an AI safety and research company and a public benefit corporation dedicated to securing its benefits and mitigating its risks. For businesses embedding these models into core workflows, the weekend outage was a reminder that performance, reliability, and vendor transparency now sit alongside model quality as strategic questions.