Skip to main content

Return to Changelog

v1.6.55

· One min read

This patch release adds chat generation control endpoints, faster document counting, and permission and citation improvements.

New features​

  • Optimized document counting: Document count operations are now significantly faster, improving performance when browsing large collections.
  • Chat generation control: Added new Python client methods (pause_chat, resume_chat, stop_chat, finish_chat) and REST endpoints to pause, resume, stop, and finish streaming chat responses.

Improvements​

  • REST API sample configuration: Updated host configuration in the sample REST API implementation to reduce connection errors.
  • Improved permission system: Group checks and permission checks are enforced when restricted mode is set, enabling the prevention of unintended access restrictions when restricted mode is disabled.
  • PDF previews: Added page numbers to PDF previews for ease of reference.
  • Citation reliability: Improved handling of unstable LLM outputs to ensure citations render correctly in more edge cases.

Bug fixes​

  • Parameter validation: Added safeguards for cases where insufficient parameters are provided.
  • Security updates: Addressed multiple CVE security vulnerabilities for platform security.
Feedback