
In v1.26.0, LobeHub adds a persistent Agent sidebar to make high-frequency Agents easier to reach during active conversations. Instead of jumping between lists, you can keep your key Agents visible and switch with less friction.
Tighter conversation workflow
This update focuses on reducing small interaction costs in everyday chat:
- Quick switching — Use keyboard shortcuts to move between Agents faster
- More chat space — When the sidebar is active, the conversation list hides to free up room for messages
- Synced pinned Agents — Your pinned Agents automatically appear in the sidebar so priority items stay in view


How to enable
This capability is currently experimental and disabled by default. To try it in self-hosted setups, enable it with:
FEATURE_FLAGS=+pin_listThe Cloud version already has this enabled. If you try it, share feedback in GitHub Discussions so we can keep improving the experience.
Why we built it
The goal of this release is straightforward: improve focus and reduce small interaction costs in everyday chat. By keeping frequently used Agents close at hand and reclaiming screen space for messages, the interface supports longer, more concentrated conversations.
Version details
Features
What's improved
- v1.25.2Update stepfun models.
- v1.24.2Update Google Model list, add gemini-1.5-flash-8b.
- v1.24.1Refactor the Google Gen AI.
- v1.23.1Add *_MODEL_LIST env to all models, update Spark model id & display name.
- v1.22.27Add bedrock claude-3.5-sonnect-v2.
- v1.22.26Fix some custom branding detail.
- v1.22.25Remove unused user tables.
- v1.22.24Support plugin flag.
- v1.22.23Improve error i18n.
- v1.22.22Improve i18n.
- v1.22.21Refactor cookie/headers to async mode.
- v1.22.20Add new claude-3.5-sonnet model.
- v1.22.19Move responsive to server utils folder.
- v1.22.17Fix dynamic import in rsc layout.
- v1.22.15Update format utils and shared layout.
- v1.22.14Update wenxin 4.0 turbo model to latest.
- v1.22.13Add Ministral model, update Together AI model list, add function call & vision.
- v1.22.12Add Llama 3.1 Nemotron 70B model & reorder some provider model list.
- v1.22.11Refactor azure ad to ms entra id.
- v1.22.9Update Fireworks AI model list.
- v1.22.8Add Yi-Lightning model.
- v1.22.7Add qwen vision model & update qwen2.5 72b to 128k for siliconcloud.
- v1.22.5Reorder github model list & updata info & add new model.
- v1.22.4Separate message slice and aiChat slice.
- v1.22.3Support multi-windows for PWA.
- v1.22.0Refactor the chat webapi.
- v1.21.14Fix artifacts render markdown.
- v1.21.13Refactor agent runtime implement of stream and ZHIPU provider.
- v1.21.12Refactor the jwt code.
- v1.21.11Refactor the backend code for better organization.
- v1.21.10Updata gpt-4o model info.
- v1.21.9Update qwen vl model to latest.
- v1.21.8Support yml in file chunk.
- v1.21.7Refactor text-to-image endpoint.
- v1.21.6Move backend api to (backend) folder group.
- v1.21.5Support shadcn in Artifacts.
- v1.21.3Move most /api to /webapi.
- v1.21.2Adjust Wenxin icon size.
- v1.20.7Update groq model list.
- v1.20.2Add zhipu glm-4-flashx model.
- v1.19.36Add llama3.2 model for openrouter provider.
- v1.19.35Add o1-preview and o1-mini model to github model provider.
- v1.19.33Update google provider model info.
- v1.19.32Add function call for taichu_llm.
- v1.19.31Add google gemini 1.5 002 series.
- v1.19.30Disable taichu2.0 functioncall & default disable taichu2.0v model.
- v1.19.29Update taichu provider info & add taichu vision model.
- v1.19.28Add function call support for Stepfun.
- v1.19.27Improve images display in chat messages.
- v1.19.26Add support function call for 360AI, left sidebar has only assistants.
- v1.19.23Add spark max-32k model.
- v1.19.21Refactor to improve branding customization.
- v1.19.16Improve i18n for discover and improve version check.
- v1.19.15Improve i18n in discover.
- v1.19.12Support webhooks for casdoor.
- v1.19.10Add qwen2.5 math and coder model for siliconcloud provider.
- v1.19.8Delete siliconflow outdated model & disable stepfun functioncall.
- v1.19.7Add siliconflow qwen2.5 model.
- v1.19.6Refactor the tts route url.
- v1.19.5Enable functioncall for stepfun models, Update qwen models.
- v1.19.4Refactor the sitemap implement.
- v1.19.2Updata qwen model info & add qwen2.5 & reorder provider list.
- v1.19.1Add mistral provider new models.
What's fixed
- v1.25.3Fix the issue of the switch assistant portal not closing.
- v1.25.1Fix modelList merge.
- v1.22.16Fix azure-ad.
- v1.22.6Fix images not go in to chat context.
- v1.22.2Allow use email as name in logto.
- v1.22.1Fix function calling issue, disable stream when using tools.
- v1.21.8Fix auto rewrite query when user message is too long.
- v1.21.6Fix txt-to-image api.
- v1.21.4Fix recharts deps in the Artifacts React Renderer.
- v1.20.3Improve delete orphan chunks when delete files.
- v1.19.33MiniMax output long content interrupted by non-existent error.
- v1.19.26Fix url config import after user state init.
- v1.19.25Add missing translations.
- v1.19.24Fix artifacts code language highlight.
- v1.19.22Fix ollama model download panel.
- v1.19.20Fix Content-Security-Policy.
- v1.19.19Casdoor webhooks providerAccountId not found.
- v1.19.17providerAccountId not exist in provider.
- v1.19.14Pin shiki@1.17.7 to fix code highlight.
- v1.19.13Try to implement better ssrf-protect.
- v1.19.11Custom model initialization not taking effect error.
- v1.19.9Fix a bug with server agent config when user not exist.
