Users report Claude quality drop

Developers who use Anthropic’s Claude for coding are reporting worse answers, more mistakes, and shakier behavior, even as Anthropic says it has been adjusting settings and shipping fixes. (venturebeat.com, anthropic.statuspage.io) The complaints accelerated in April 2026, after Anthropic released Claude Opus 4.6 on February 5 with claims of stronger coding, debugging, and longer agentic work across large codebases. Anthropic’s status page also logged a Claude.ai and Claude Code login outage on April 13 from 15:31 to 16:19 Coordinated Universal Time. (anthropic.com, anthropic.statuspage.io) Claude Code is Anthropic’s coding agent, which means it can read a repository, edit files, run tests, and return code instead of just suggesting snippets. Anthropic markets it as a system that can work across files and deliver committed code, so users notice quickly when it starts losing context or producing weaker patches. (anthropic.com) The current dispute is not just about one bad response. VentureBeat reported that users and some internal metrics flagged a decline in code quality and general performance, while Anthropic leaders argued that tuning different settings for different user segments can change how the product feels even if the underlying model weights stay the same. (venturebeat.com) The Register found a similar pattern in public complaints around Claude Code, but also noted that some issue traffic may be noisy because artificial intelligence systems themselves are generating bug reports and commentary. That leaves two overlapping problems for Anthropic: real user frustration and a harder job separating signal from spam. (theregister.com) Anthropic has spent the past year pushing Claude Code deeper into developer workflows. In September 2025, it added a Visual Studio Code extension, a new terminal interface, and checkpoints for more autonomous operation, and in August 2025 it expanded Claude Code on Team and Enterprise plans with premium seats and spend controls. (anthropic.com, anthropic.com) That product push raises the stakes for any perceived drop in quality, because Claude Code is no longer a side feature for many paying teams. Anthropic says Team and Enterprise administrators can assign premium seats, monitor usage analytics, and allow extra usage at standard application programming interface rates. (anthropic.com) At the same time, Anthropic appears to be preparing a broader redesign. TestingCatalog reported on April 13 that Anthropic is testing a Claude Code desktop upgrade called Project Epitaxy with new panels, multi-repository support, and a “Coordinator Mode” aimed at more complex software work. (testingcatalog.com) Anthropic’s own recent messaging still frames Claude Opus 4.6 as its strongest model for coding and professional work, with a 1 million token context window in beta and better code review and debugging. The gap between that promise and the latest user reports is what turned routine product grumbling into a broader argument over whether Claude has been “nerfed.” (anthropic.com, venturebeat.com) For now, the public record shows three things at once: a fresh outage on April 13, a visible wave of user complaints about coding quality, and an unreleased Claude Code overhaul still in testing. Anthropic is trying to reassure developers while asking them to trust a tool that many of them say no longer behaves like it did a few weeks ago. (anthropic.statuspage.io, venturebeat.com, testingcatalog.com)