Anthropic has announced that it discovered a problem leading to higher error rates, mainly affecting its Claude Opus 4.7 and Sonnet 4.6 models. This comes after the company had conducted postmortems on recent updates to its Claude models, which highlighted issues with coding quality and error handling in the 4.6 and 4.7 versions. Anthropic is actively working on improving the reliability of its Claude series, with ongoing efforts to enhance agentic capabilities and ensure better error recovery during complex tasks.

Sonnet: Sonnet is Anthropic’s balanced-performance model line, with the 4.6 version serving as a capable option for iterative development, document analysis, and workflow automation. It offers a practical tradeoff between intelligence and efficiency compared to the more resource-intensive Opus models. The news highlights how this specific Sonnet version is among those experiencing increased error rates from the identified issue.
Anthropic: Anthropic is an AI company specializing in developing advanced large language models with a strong emphasis on safety and reliability. It recently launched Claude Opus 4.7 in April 2026, building on prior versions with enhancements for complex agentic workflows and coding tasks. In the current news, Anthropic publicly reported identifying a specific issue causing elevated error rates primarily in its Opus 4.7 and Sonnet 4.6 models.
Claude Opus: Claude Opus refers to Anthropic’s flagship high-capability model series, with the 4.7 version released in mid-April 2026 as an upgrade focused on better performance in multi-step reasoning and tool use. It is designed for demanding professional and enterprise applications such as long-running agent pipelines. The reported problem in the news directly impacts error rates in this latest Opus iteration alongside related models.

Model Reliability: Anthropic has conducted and published postmortems on recent changes to its Claude models that affected coding quality and error handling across the 4.6 and 4.7 versions.
Product Development: The company continues to iterate on its Claude family with a focus on agentic capabilities and graceful error recovery in complex tasks.