An update on recent Claude Code quality reports

Get the developer newsletter

Product updates, how-tos, community spotlights, and more. Delivered monthly to your inbox.

Over the past month, we’ve been looking into reports that Claude’s responses have worsened for some users. We’ve traced these reports to three separate changes that affected Claude Code, the Claude Agent SDK, and Claude Cowork. The API was not impacted.

All three issues have now been resolved as of April 20 (v2.1.116).

In this post, we explain what we found, what we fixed, and what we’ll do differently to ensure similar issues are much less likely to happen again.

We take reports about degradation very seriously. We never intentionally degrade our models, and we were able to immediately confirm that our API and inference layer were unaffected.

After investigation, we identified three different issues:

1. Reasoning Effort Defaults

When we released Opus 4.6 in Claude Code in February, we set the default reasoning effort to high. Soon after, we received user feedback that Claude Opus 4.6 in high effort mode would occasionally think for too long, causing the UI to appear frozen and leading to disproportionate latency and token usage for those users.

In our internal evals and testing, medium effort achieved slightly lower intelligence with significantly less latency for the majority of tasks. However, users began reporting that Claude Code felt less intelligent. After hearing feedback from more customers, we reversed this decision on April 7. All users now default to xhigh effort for Opus 4.7, and high effort for all other models.

2. Thinking History and Prompt Caching Bug

On March 26, we shipped an efficiency improvement to prompt caching. The design intended to clear old thinking sections only if a session had been idle for more than an hour to reduce costs. However, the implementation had a bug: it cleared thinking history on every turn for the rest of the session once the idle threshold was crossed.

This caused Claude to lose memory of why it made previous choices, leading to forgetfulness, repetition, and odd tool choices. It also caused cache misses, draining usage limits faster. We fixed this bug on April 10 in v2.1.101.

3. System Prompt Verbosity Constraints

To manage the verbosity of Opus 4.7, we added length limits to the system prompt (e.g., keeping text between tool calls to ≤25 words). After further investigation, we found these constraints caused a 3% drop in intelligence for both Opus 4.6 and 4.7. We immediately reverted the prompt change.

Moving Forward

This isn’t the experience users should expect from Claude Code. As of April 23, we’re resetting usage limits for all subscribers. We are also landing support for additional repositories as context for code reviews to prevent similar regressions in the future.