History
Share of Anthropic internal Claude Code sessions on the most open-ended tasks judged successful in May 2026 (%)
Structured and assessed
First pass. Read the Anthropic Institute report in full. Completed the definition (Claude-judge success, most open-ended tier, internal Claude Code sessions, May 2026). Recorded the reported figure (76%) as an estimate under one argument from the company's testimony, rated likely. Estimate 76% with 90% interval 71-80, reflecting unpublished data and fluctuating series. Recorded provenance reading as evidence not shown. Importance 0.3, contestation 0.25: the value is not disputed; its meaning is. No dependents to notify.
Claim entered the graph