📢 LibreChat Platform Upgrade - v0.8.4 Releasing Tonight @ 5:00 PM
Hi everyone,
We will be releasing version 0.8.4 of our AI Chat platform to production tonight at 5:00 PM, upgrading from version 0.8.1. This update brings several improvements to enhance your experience:
🎯 What's New for You:
✨ New Model Provider Icons
- Each AI model now displays its provider icon for quick visual identification
- OpenAI icon appears for all GPT models (GPT-4o, GPT-4o Mini, GPT-4.1, GPT-4.1 Mini, GPT-5)
- Google icon appears for Gemini models (Gemini 2.5 Pro, Gemini 2.5 Flash)
- Anthropic icon appears for Claude Sonnet 4.5
- Icons are visible both in the model selection menu and in your conversation header
- Makes it easier to see at a glance which AI provider you're chatting with
📎 File & Attachment Improvements
- More file types supported — you can now upload OpenDocument files (
.odt,.ods) and Excel spreadsheets (.xlsx,.xls) and have their content read and used directly in your conversations — no need to convert files before uploading - File Search is fixed so it correctly recognizes your uploaded files when you or your agent uses search across documents — previously some searches could fail or be blocked even though you had uploaded the file
- Image uploads now work with Claude Sonnet — images you attach are now correctly sent to Claude; previously they were being silently ignored
🧠 Prompt Catalog Integration (New!)
- Paychex AI Knowledge Base now connects directly to LibreChat
- You can insert prompts from the Knowledge Base into a new conversations without copy-pasting
- Helps standardize prompts across teams and reuse known-good prompts
🤝 Agent Improvements
- Agents now display support contact information on the landing page
- Agent conversations automatically get descriptive titles (instead of defaulting to "New Chat")
User Experience Enhancements
- Smoother, more responsive conversations with all available models
- Improved model selection interface - easier to switch between models
- Enhanced parameter controls for fine-tuning your responses
- Better bookmark management for saving your favorite conversations
- Improved prompt library functionality - easier access to saved prompts
Performance & Stability
- Faster response times across all models
- More reliable conversation handling, especially for longer discussions
- Improved error recovery when issues occur
- Better handling of conversation context and history
Interface Refinements
- Cleaner, more intuitive layout
- Smoother transitions and interactions
- Better visual feedback during conversations
🔧 Behind the Scenes
- Updated security measures to keep your data safe
- Infrastructure improvements for better reliability
- Bug fixes and performance optimizations, including the File Search authorization alignment described above (for teams using document search with agents)
📝 What Hasn't Changed:
- All your existing conversations, bookmarks, and prompts are preserved
- Your available models remain the same
- The interface you're familiar with works just like before
Need Help?
If you notice anything unusual or have questions, please reach out to the support team.
Thank you for your patience during the upgrade!
What's Changed
- fix: Tavily rate limiting errors by adding instruction to avoid research endpoint by @gmarcelino-paychex in #126
- fix(tavily): updated server instructions to address tavily issues by @gmarcelino-paychex in #127
- Update cxone-github-action.yml by @paychex-joser in #129
- feat: add /refresh to clear models cache by @gmarcelino-paychex in #128
- AIA-2787 LangGraph Agents Title Summaries in LibreChat by @paychex-gcarlson in #130
- feat: add agent support contact information in Landing component by @gmarcelino-paychex in #132
- Add identification-specific LangGraph endpoint configs by @paychex-inder in #133
- Fetch identification models in n1 by @paychex-inder in #136
- fix: workaround for parallel Claude tool call SSE index issue by @gmarcelino-paychex in #134
- feat: enable langgraph identification fetch in prod by @paychex-inder in #138
- Feature/aia 2207 merge upstream v0.8.4 by @paychex-joser in #137
- feature/AIA-1936 add icons for each ai model provider (Updated) by @paychex-joser in #140
- fix: reduce MCP SSE transient error log noise and prevent reconnectio… by @paychex-joser in #141
- fix: Convert image format from OpenAI to Anthropic for custom Claude endpoints by @gmarcelino-paychex in #142
- fix: handle RAG context 404 gracefully in createContextHandlers by @paychex-joser in #143
- feature/AIA-2203- Add Prompt Catalog Insert to LibreChat by @gmarcelino-paychex in #144
Full Changelog: payx-20260317-lc-v0.8.1...payx-20260429-lc-v0.8.4