AnythingLLM

Introduces 'Magic Features' bringing on-device AI dictation, text actions, and autocomplete across the OS, alongside the introduction of a Pro tier and various bug fixes.

Overhauled the Meeting Assistant for better performance and GPU support, introduced speaker identification, added chat export options, and implemented various UI and API improvements.

Introduces Cerebras provider, expanded STT/TTS support, markdown web-scraping, and changes tool calling to opt-out by default.

Introduces a hybrid Model Router for local and cloud AI, scheduled agent jobs, automatic memories for personalization, and an agent survey tool.

Introduces streamed document embedding for better performance, new agent integrations for Gmail, Outlook, and Google Calendar, and expanded multi-language support for PDF generation.

Introduces automatic tool calling, intelligent tool selection, filesystem and document generation agents, and a Telegram bot integration.

This release introduces significant UI improvements to the chat interface, enhances Agent capabilities with native tool calling and better citations, and adds Perplexity Search API support.

This release introduces a homepage redesign, native tool calling for local LLM providers, Lemonade by AMD integration, and various UI and stability fixes.

Introduces a new OS-level desktop overlay for seamless application ingestion, alongside SambaNova and PrivateModeAI integrations, UI improvements, and various bug fixes.

Introduces the AnythingLLM Desktop Assistant for meeting recording, launches the Android mobile app, adds Cohere agent support, and migrates Azure OpenAI to the v1 API.

Introduces PWA support, SerpApi web search, Z.ai provider, and expanded YouTube and Paperless ngx integration alongside various LLM provider updates.

Introduces a major agent overhaul featuring real-time tool call streaming across all providers and the ability for agents to ingest files directly from the web.

v1.8.5 introduces an overhauled document chat system for better context handling, experimental Android mobile app support, and new providers including Moonshot AI and Exa search.

This update introduces workspace and thread searching in the sidebar, SQL preflight connection validation, and several UI improvements to codeblocks and message handling.

Introduces a built-in browser tool for scraping gated content, allows publishing agent flows and prompts to the Community Hub, and includes various bug fixes and new translations.

Introduces in-chat model swapping, system prompt version tracking, PGvector support, and various data connector enhancements including Obsidian and GitLab wikis.

Introduces MCP tool use via API, adds a DrupalWiki collector, enhances TTS/STT automation, and updates various model supports and translations.

Introduces full Model Context Protocol (MCP) compatibility, a new onboarding experience, a redesigned chat home page, and various UI improvements and bug fixes.

Introduces workspace system prompt variables, NVIDIA NIM support for desktop, and various API and provider enhancements.

This release introduces workspace reordering, a folder removal API endpoint, Danish translations, and various UI and database fixes.

This update introduces the Anthropic /models endpoint, multi-language OCR support, Claude 3.7, and various UI enhancements including new agent animations and layout configurations.

Introduces a no-code agent skill builder and OCR scanning for PDFs and images, alongside DeepSeek API reasoning support and Turkish language localization.

Version 1.7.3 introduces UI support for thinking models and improved agent calling interfaces, alongside expanded LLM provider support and various security and performance fixes.

Introduces the AnythingLLM Community Hub for sharing system prompts and agent skills, adds official dark and light modes, and includes various LLM integrations and UI improvements.

This release adds support for OpenAI o1, Grok/XAI, and LiteLLM agents, introduces daily message limits per user, and includes various bug fixes and UI improvements.

This release introduces support for @agent custom skills, copy-paste functionality for images and files, Tavily web search, XLSX file support, and expanded LLM integrations including DeepSeek and Fireworks AI.

Introduces custom agent skills, a new browser extension for web embedding, and enables agent skill execution via the developer API.

This update includes various bug fixes, UI improvements, support for Go files and Hebrew language, and enhancements to agent context and model window limits.

Introduces PiperTTS for in-browser text-to-speech, Portuguese and German translations, LaTeX support, and various bug fixes and API improvements.

Added Gemini Pro 1.5 support with 2M context, Italian translations, Ollama performance mode, and improved OpenRouter exception handling.

Introduces multimodality support and drag-and-drop document uploads, alongside various API updates and UI fixes.

Initial formal release introducing comprehensive support for multiple LLM providers, various vector databases, multi-user instances, a developer API, and a complete UI overhaul.