Skip to content

Lollms 2.1.0 Restart

Latest
Latest

Choose a tag to compare

View all tags
@ParisNeo ParisNeo released this 28 Dec 03:18
· 290 commits to main since this release
v2.1.0
a9b21a0

This version marks a complete architectural evolution from the legacy lollms_webui (v1.16.0 "Twins"). LoLLMs is now a full-scale, multi-user AI orchestration platform.

Added

  • Multi-User Foundation: High-performance SQLite/SQLAlchemy backend replacing legacy YAML files. Supports secure authentication (JWT), user registration modes, and persistent session management.
  • LoLLMs Client Integration: Core logic moved to the unified lollms_client library for faster, more reliable model interactions.
  • SafeStore (RAG) Studio: Full implementation of Retrieval Augmented Generation. Create persistent Data Stores, index massive document libraries, and perform semantic search across your local knowledge.
  • AI Social Feed: A community-driven feed allowing users to share posts and interact. Featuring @lollms—an autonomous bot capable of researching topics via Google, DuckDuckGo, ArXiv, and real-time Web Scraping.
  • Messaging & Groups: Direct messaging system between users and the ability to organize chats into nested, hierarchical groups.
  • Specialized Studios:
    • Notebook Studio: Document-based writing with integrated AI editing and long-form research.
    • Image Studio: Advanced control over TTI (Text-to-Image) with prompt history and generation workflows.
    • Voice Studio: Design and test high-quality AI voices using the new TTS (Text-to-Speech) pipeline.
  • MCP Bridge: Native support for Model Context Protocol (MCP), allowing any model to use external tools like calculators, search engines, and local file systems.
  • Admin Command Center: Real-time hardware monitoring (GPU/CPU/RAM), user provisioning (SCIM), Single Sign-On (SSO) configuration, and global service management.
  • Multi-Worker Synchronization: New Communication Hub (ComHub) for seamless event broadcasting across high-concurrency server clusters.
  • Auto-Posting Service: Configurable bot scheduler that uses knowledge bases to generate engaging community content autonomously.

Changed

  • UI Redesign: Complete shift to a modern, responsive Vue 3 + Tailwind CSS interface with beginner to expert UI levels.
  • API Standard: Fully compatible OpenAI v1 and Ollama v1 endpoints serving local models to external applications.

Fixed

  • Resolved many concurrency issues inherent in the single-user architecture of previous versions.
  • Improved model loading performance with intelligent process caching.
Assets 2
Loading