Microsoft researchers lift AI-agent memory pass rate from 39% to 73%

Verifying
  • Microsoft researchers introduced environment-probing curation for AI agents' long-term memories.
  • CLBench pass rates rose from 39% to 73%, while cost fell from $3.38 to $1.68.
  • Testing covered six APEX scenarios and multiple foundation models, including Sonnet 4.6 and Opus 4.7.
The information on this website is generated using AI and we cannot guarantee its accuracy. Please use it as reference information only.