OSArchURL
Macx64https://cdn.anythingllm.com/legacy/1.14.0/AnythingLLMDesktop.dmg
MacARM64https://cdn.anythingllm.com/legacy/1.14.0/AnythingLLMDesktop-Silicon.dmg
Windowsx64https://cdn.anythingllm.com/legacy/1.14.0/AnythingLLMDesktop.exe
WindowsARM64https://cdn.anythingllm.com/legacy/1.14.0/AnythingLLMDesktop-Arm64.exe
Linuxx64https://cdn.anythingllm.com/legacy/1.14.0/AnythingLLMDesktop.AppImage
LinuxARM64https://cdn.anythingllm.com/legacy/1.14.0/AnythingLLMDesktop-Arm64.AppImage

Key Behavior Changes

  • Default Chat Thread: Automatically terminated upon creation of a new thread. Existing chats on the default thread remain accessible. Default thread is hidden for new workspaces or workspaces with no active chats on the default thread.
  • Tool Calling: Now opt-out by default for all model providers.
  • Web-Scraping: Converts targets to Markdown to minimize context usage.
  • Summary Tool: Requires user confirmation before executing longer summaries.
  • System Variables: Supports 24-hour time formatting.
  • Windows Uninstaller: Includes a checkbox option to purge all user data.

Provider Additions & Updates

  • LLM/API: Cerebras provider added. GenericOAI provider improved.
  • STT (Speech-to-Text): Deepgram, GenericOAI, Lemonade, and OpenAI supported.
  • TTS (Text-to-Speech): KokoroTTS supported.

Bug Fixes

  • Resolved agent context limit detection issues.
  • Fixed SEARXNG double encoding.
  • Implemented global timeouts for all fetch requests.
  • Added escaping for illegal XML in Word documents (.docx).
  • Resolved system tray issues when launching in background or toggling the Desktop Assistant.
  • Improved LaTeX rendering.