Download Links (v1.13.0)
| OS | Arch | Download URL |
|---|---|---|
| macOS | x64 | https://cdn.anythingllm.com/legacy/1.13.0/AnythingLLMDesktop.dmg |
| macOS | ARM64 | https://cdn.anythingllm.com/legacy/1.13.0/AnythingLLMDesktop-Silicon.dmg |
| Windows | x64 | https://cdn.anythingllm.com/legacy/1.13.0/AnythingLLMDesktop.exe |
| Windows | ARM64 | https://cdn.anythingllm.com/legacy/1.13.0/AnythingLLMDesktop-Arm64.exe |
| Linux | x64 | https://cdn.anythingllm.com/legacy/1.13.0/AnythingLLMDesktop.AppImage |
| Linux | ARM64 | https://cdn.anythingllm.com/legacy/1.13.0/AnythingLLMDesktop-Arm64.AppImage |
Core Feature Configurations
1. Model Router (Hybrid AI)
Enables rule-based routing of chat messages to local models (e.g., Ollama, LM Studio) or cloud providers (e.g., OpenAI, Anthropic, Google).
- Calculated Rules: Trigger routing based on keywords, token count thresholds, time of day, or image attachments.
- LLM-Classified Rules: Route messages based on intent classification evaluated by an LLM.
- Sticky Routing: Maintains connection to the same model within a conversation thread to prevent continuous switching.
- Reference:
/model-router/setup
2. Scheduled Jobs (Background Agents)
Executes agent tasks on a recurring schedule with access to tools (web search, document analysis, custom skills, MCP integrations).
- Trigger Mechanism: Visual Cron Builder or cron schedules.
- Execution Logs: Contains agent reasoning, tool calls, generated files, and final responses.
- Notifications: Push notifications trigger upon job completion.
- Reference:
/scheduled-jobs/getting-started
3. Automatic Memories & Personalization
Background extraction of user information injected into system prompts.
- Global Memories: Instance-wide preferences (e.g., user name, global settings).
- Workspace Memories: Scope-limited to specific workspaces (e.g., project data).
- Reference:
/features/memories
4. Agent Surveys
Allows agents to prompt the user with clarifying questions before executing tasks.
- Configuration: Disabled by default. Toggle on via Agent Settings.
- Reference:
/features/agent-surveys
Changelog & Improvements
Features & Integrations
- Chat UI: Updated “Auto” mode label to “Agent” (
/features/chat-modes#available-chat-modes). - Tool Management: Direct access to view and manage active tools from the chat window.
- Web Search: Added Baidu search engine support.
- Providers: Added MiniMax support; added reasoning support for LMStudio/Lemonade.
- API: Added endpoint to pull generated documents.
Bug Fixes & Security
- Fixed chat history cleanup bug requiring
/resetto be executed twice. - Fixed markdown tag pronunciation in Text-to-Speech (TTS).
- Fixed “Auto speak” execution on new incoming messages.
- Fixed Deepseek v4 thought reinjection errors during chats.
- Fixed Gemini parallel tool calling errors and optimized Gemini error reporting.
- Fixed Cyrillic font fallback rendering.
- Fixed Community Hub CSS layout clipping on narrow viewports.
- Implemented security updates and enhanced UI tooltips.