In March, LobeHub significantly enhanced its search infrastructure and introduced agent document capabilities, laying the groundwork for smarter knowledge retrieval.

🌟 Key Updates

• 🔍 BM25 search indexes: Added BM25 indexes with ICU tokenizer for optimized full-text search
• 📄 Agent documents: Introduced the agent_documents table for agent-level knowledge storage
• 🗄️ pg_search extension: Enabled the pg_search PostgreSQL extension for advanced search capabilities
• 📝 Topic descriptions: Added description column to the topics table for better topic organization
• 🔑 API key security: Added API key hash column for enhanced security

💫 Experience Improvements

Fixed changelog auto-generation in release workflow, corrected stable renderer tar source path, and resolved market M2M token registration for trust client scenarios.

In February, LobeHub focused on model runtime enhancements, authentication reliability, and polishing the overall user experience across platforms.

🌟 Key Updates

• 🤖 Claude Opus 4.6 on Bedrock: Added Claude Opus 4.6 support for AWS Bedrock runtime
• 📓 Notebook tool: Registered Notebook tool in server runtime with improved system prompts
• 🔗 OpenAI Responses API: Added end-user info support on OpenAI Responses API calls
• 🔐 Auth improvements: Fixed Microsoft authentication, improved OIDC provider account linking, and enhanced Feishu SSO
• 📱 Mobile enhancements: Enabled vertical scrolling for topic list on mobile, fixed multimodal image rendering
• 🏗️ Runtime refactoring: Extracted Anthropic factory and converted Moonshot to RouterRuntime

💫 Experience Improvements

Improved tasks display, enhanced local-system tool implementation, fixed PDF parsing in Docker, fixed editor content loss on send error, added custom avatars for group chat sidebar, and showed notifications for file upload storage limit errors.

January marks the landmark release of LobeHub v2.0, introducing powerful multi-agent group chat capabilities, refined model settings, and a streamlined authentication experience.

What's New

• A major version upgrade with redesigned architecture and enhanced features
• Multi-Agent Collaboration: Bring multiple specialized agents into one conversation. They debate, reason, and solve complex problems together—faster and smarter.
• Agent Builder: Describe what you want, and LobeHub builds the complete agent—skills, behavior, tools, and personality. No setup required.
• Pages: write, read and organize documents with Lobe AI
• Memory: Your agents remember your preferences, style, goals, and past projects—delivering uniquely personalized assistance that gets better over time.
• New Knowledge Base: Use folders to organize your knowledge & resource
• Marketplace: Publish, adopt, or remix agents in a thriving community where intelligence grows together.

Improvement

• Enhanced model settings: New ExtendParamsTypeSchema for more flexible model configuration
• Model updates: Updated Kimi K2.5 and Qwen3 Max Thinking models, plus Gemini 2.5 streaming fixes

In November, LobeHub continued to improve model support and user experience by adding multiple AI providers and enhancing knowledge base capabilities.

🌟 Key Updates

• 🔌 MCP cloud endpoints: supports integrating marketplace cloud-endpoint MCPs to expand the tool ecosystem
• 🤖 New providers: adds support for multiple AI providers including ZenMux, Nano Banana Pro, Qiniu Cloud, and more
• 📚 Knowledge base enhancements: supports creating pages, improves file management, and enhances RAG search experience
• 🎨 Image generation: adds support for more image models and improves image generation configuration
• 🔐 Auth improvements: improves OIDC authentication flow and desktop login experience
• 💬 Conversation improvements: supports topic hyperlinks and improves message editing and deletion

💫 Experience Improvements

Improves topic list interactions, enhances tool call display, refines rich text editing, optimizes token usage statistics animations, and improves model selector sorting.

From October to November, LobeHub integrated ComfyUI workflows, added support for multiple AI providers and models, and continued to improve the knowledge base and overall user experience.

🌟 Key Updates

• 🎨 ComfyUI integration: supports integrating ComfyUI workflows
• 🤖 New providers: adds support for providers like Cerebras and CometAPI
• 📄 PDF export: export conversations as PDF
• 🗂️ Knowledge base improvements: adds a waterfall layout and automatically unzips files on upload
• 🖼️ Image generation expansion: supports SiliconFlow and Hunyuan Text-to-Image 3 image generation services

📊 Model Library Updates

Adds models such as Claude Haiku 4.5, GPT-5 Pro, MiniMax-M2, and Imagen 4 for Vertex AI.

💫 Experience Improvements

Improves rich text link rendering, enhances search experience, supports disabling rich text editing, adds hotkeys for delete and regenerate, and improves update notifications.