"Anthropic adjusted Claude Code's default reasoning effort level from high to medium on March 4, hoping to reduce latency. This change was deemed a misstep, leading to user dissatisfaction."
"A bug introduced on March 26 cleared cached session data with every prompt-response cycle, negatively affecting the speed and cost-effectiveness of sequential API calls for users."
"Users expressed frustration over the perceived decline in Claude's output quality during March and April, prompting Anthropic to revert the reasoning effort level back to high."
Anthropic's investigation revealed that changes made to Claude's service in March and April negatively impacted response quality. The default reasoning effort level was reduced from high to medium, which increased latency. A bug introduced on March 26 cleared cached session data, further degrading user experience. Users reported dissatisfaction with the output quality, prompting Anthropic to revert the reasoning effort level change. The latest build of Claude Code now defaults to a higher reasoning effort to improve performance.
Read at Theregister
Unable to calculate read time
Collection
[
|
...
]