
From July through August, LobeHub shipped AI image generation, expanded identity and access options, and kept refining the desktop experience. The common thread is practical capability you can use immediately鈥攎ore ways to create visuals, easier connections to existing systems, and smoother day-to-day workflows on desktop.
Generate images your way
LobeHub now supports AI image generation through Google Imagen, Qwen, Zhipu CogView4, and MiniMax. Pick the provider that matches your quality, speed, and budget preferences without changing your overall workflow.
Use this to:
- Generate visuals directly in chat without switching tools
- Compare outputs across providers for the same prompt
- Route image requests based on your current project needs
Connect to existing identity systems
Authentication now extends to Amazon Cognito, Google SSO, and Okta. If your organization already runs on these systems, connecting LobeHub is now a configuration step rather than a migration project.
This release also adds authentication support for Streamable HTTP MCP servers, and the API Key management interface is now fully functional for creating, managing, and rotating keys.
Desktop and model updates
Desktop improvements include network proxy configuration, custom hotkey support, OAuth flow refinements, and remote chat capabilities. These changes make the desktop app more adaptable across different network environments and team setups.
The model library expands with Claude Opus 4.1, Grok-4, Kimi K2, and Ollama gpt-oss, plus updates to Gemini 2.5 Flash-Lite GA, Hunyuan A13B thinking, and Doubao reasoning models.
Improvements and fixes
- Added desktop notifications
- Refined settings window layout
- Improved multi-monitor behavior
- Enhanced MCP plugin invocation and rendering
- Fixed Gemini Artifacts line-break issues
Version details
Features
- v1.109.0Support gpt-oss in ollama provider.
- v1.108.0Support 302ai provider.
- v1.107.0Support aihubmix provider.
- v1.106.0Add support for Okta Authentication.
- v1.105.0Implement API Key management functionality.
- v1.104.0Support custom hotkey on desktop.
- v1.103.0Add Qwen image generation capabilities.
- v1.102.0Add image generation capabilities using Google AI Imagen API.
- v1.101.0Add zhipu cogview4.
- v1.100.0Refactor desktop oauth and use JWTs token to support remote chat.
- v1.99.0support AI Image.
- v1.98.0Add network proxy for desktop.
What's improved
- v1.109.1Add Claude Opus 4.1 model, update i18n.
- v1.108.1Update mask style.
- v1.107.5Update models.
- v1.107.2Move types to separate package.
- v1.107.1Update i18n.
- v1.106.8Support SenseNova V6.5 models.
- v1.106.7Update Aliyun Bailian models.
- v1.106.5Improve mcp plugin calling and display.
- v1.106.4Update i18n.
- v1.106.3Add volcengine kimi-k2 model, Add Zhipu GLM-4.5 models.
- v1.106.1Support Minimax T2I models.
- v1.105.6Open new topic by tap Just Chat again.
- v1.105.2Add more OpenAI SDK Text2Image providers, update i18n.
- v1.105.1Support more Text2Image from Qwen.
- v1.104.5Fix setting window layout when in desktop was disappear.
- v1.104.4Fix setting window layout size, update i18n.
- v1.104.3Add Gemini 2.5 Flash-Lite GA model.
- v1.103.2Add cached token count to usage of GoogleAI and VertexAI, fix desktop titlebar style in window, fix sub topic width in md responsive.
- v1.103.1Update i18n.
- v1.102.4Update tray icon.
- v1.102.2Add notification for desktop.
- v1.102.1Modal list header sticky style.
- v1.101.2Fix lobehub provider /chat in desktop.
- v1.98.2Update i18n.
- v1.98.1Fix discover translation.
- v1.97.17Support Hunyuan A13B thinking model.
- v1.97.16Update i18n.
- v1.97.14Add Kimi K2 model.
- v1.97.13Support new Doubao thinking models, update i18n.
- v1.97.11Open new topic by tap Just Chat again.
- v1.97.10Update i18n.
What's fixed
- v1.109.1Fix ollama model output without thinking.
- v1.108.2Provider config checker uses outdated API key.
- v1.108.1Fix remote avatar broken in desktop.
- v1.107.6Break line for Gemini Artifacts.
- v1.107.4When s3 files not exist , global files should delete.
- v1.107.3Aihubmix provider request headers.
- v1.106.6Fix oidc oauth callback pages 404.
- v1.106.4Fix mcp calling missing array content.
- v1.106.3Moonshot assistant messages must not be empty.
- v1.106.2Fix desktop auth redirect url error.
- v1.105.5Reorder AppTheme and Locale to fix modal i18n.
- v1.105.4Revert jose to ^5 to fix auth issue on desktop.
- v1.105.3Fix subscription plan tag display.
- v1.104.2Fix update hotkey invalid when input mod in desktop.
- v1.104.1Update convertUsage to handle XAI provider and adjust OpenAIStream to pass provider.
- v1.103.2Fix chat stream in desktop and update shortcut.
- v1.102.3Remove debug logging from ModelRuntime and async caller.
- v1.101.1Try fix authorization code exchange & pin next-auto to beta.29.
- v1.101.0Some ai image bugs.
- v1.100.2Fix webapi proxy with clerk.
- v1.100.1Use server env config image models.
- v1.99.6Desktop local db can't upload image.
- v1.99.5Fix page error when url is not defined in web search plugin.
- v1.99.4Fix apikey issue on server log.
- v1.99.3Chat model list should not show image model.
- v1.99.2Some ai image generation feedback issues.
- v1.97.15Add vision support to Grok 4.
- v1.97.14Revert "馃拕 style: Open new topic by tap Just Chat again".
- v1.97.12Grok-4 reasoning model universal matching.
