LibreChat 2025 Roadmap
The 2025 roadmap for LibreChat: upcoming features, partnerships, and more.
LibreChat Turns 2! 🎉
I’m excited to announce that LibreChat is now officially 2 years old, making it the longest-running, active AI Chat UI! Since being the first major open-source offering following ChatGPT’s release, we’ve come a long way and accomplished so much together.
2024 Achievements
Before diving into our plans for 2025, let’s celebrate what we’ve accomplished in 2024:
- ✨ Started January 2024 with 4,000 GitHub Stars, now at 22,200+
- 📈 Reached Top 25 on GitHub Trending for All Languages over 15 times, including #1
- ⬇️ Over 3 million container registry pull downs
- 👐 Top-rated AI App for Digital Accessibility thanks to partnership with Harvard University
- 🏢 Leading Open-source AI Chat UI of choice for Large Orgs, Small Businesses, and Higher Education
As well as key feature updates:
- Agentic Tooling for all Supported Providers via LibreChat Agents
- Code Interpreter API for secure, sandboxed Execution in Python, Node.js (JS/TS), Go, C/C++, Java, PHP, Rust, Rscript, and Fortran
- Enhanced OpenAI Compatibility via Custom Endpoints, including Local LLM Integrations
- Artifacts for Generative UI (inspired by claude.ai)
- Improved Chat UI and Backend Optimizations
- Multiple Response Streaming (experimental)
- Speech-to-Text and Text-to-Speech
- In-line Text Commands for Prompts and Agent/Model switching
- 18next Internationalization Support
- Local RAG for Extended File Support
- Temporary Chats
The 2025 Roadmap
Note: for all planned/proposed items as they are being worked on, other items, especially bug fixes, may be addressed as needed.
Upcoming Features
These features will release within the next month, as part of v0.7.8 and/or v0.7.9, with some already in active development.
- Add ability to upload files as text:
- Similar to claude.ai, whatever can be parsed from the file as text goes straight into your prompt
- After upload, the file’s text is treated as an attachment and can be “sourced,” that is attached/detached, whenever needed in the conversation and/or agent, just as existing files
- Add ability to send files to provider (#4616):
- Google supports “Video Understanding” & “Document Understanding”
- Anthropic supports PDFs
- ”Grounding” Google Search for Gemini models
librechat.yaml: Specify vision models per provider or for all providers, as well as the default vision model (#1634)librechat.yaml: Custom token rates for Endpoints (#1633)librechat.yaml: Custom Welcome Message for Landing (#2967)- Chat UI Refresh, with streamlined features
- Modernized Landing Screen with Tool toggles for a simpler, starting chat UX
- Update Messages UI for a modern look/experience
- Experimental: Mixture-of-Agents
- Ability to create custom “Chain-of-thought” agents which leverage a “Mixture-of-Agents” to help optimize LLM performance and cost
- Note: This is upcoming as most of the work for this has already been prototyped, and simply requires UI/UX polish
Deadline: March 15th, 2025
Focus Item: Admin Panel
Work on this year’s focus item will start once the upcoming/short-term items are completed and/or nearing completion. This is a big project, where a lot of resources will go into in hopes of accelerating the release much sooner than the projected deadline.
The Admin panel will be a separate, optional app that acts as a Content Management System (CMS) for LibreChat, including management of usage, chats, users, settings, and access controls.
The core vision is to provide a Graphical User Interface for configuration/security related settings from the environment file, as well as providing all present/future librechat.yaml configurations an interface, while also extending the ability to apply these configurations at a role and even user level.
More details will be released as the project nears completion.
Deadline: June 6th, 2025
Planned Items
Planned items are effectively on the docket based on community interest but without a current estimation/deadline. They are also not listed in any particular order, and may be completed at any point this year.
- OAuth Client Credentials Flow for Actions (Agents/Assistants)
- OAuth Authorization flow is already implemented.
- OpenAI-compatible API for LibreChat Agents
- Add MCP Resources for wider, ad-hoc data integrations
- Sources UI, to better understand what files/images are included in the active conversation
- Side Panel UI Refresh, for less obtrusion and more intuitiveness
- Dynamic prompt variables for Agent instructions (#5007)
- User-specific memories (#5494)
- Extension of Prompt Library features to Agent/Assistant Conversation Starters (#5053)
- Favoriting and Custom Ordering of Prompt Library Entries (#4875)
- Keyboard shortcuts for Quick-access/navigation (#3664)
- Enhancing Artifacts to support markdown/notes, similar to ChatGPT’s canvas (#3995)
- Extending Artfiacts code-editing to also support the LibreChat Code Interpreter API, in order to execute Python, Rust, Go, C++, and more, as artifacts
- Human-in-the-loop for tool usage (allow user to approve a tool when used) (#5580)
- Creation of teams, groups, workspaces (#4730)
- Deep research tool
- Better search integration, including UI elements for sources/citations (#4615)
- Chat rating for feedback (#901)
- Info/metadata for individual messages and conversations (#1241, #1215, #5199)
- Includes projected cost, timestamp, etc.
- Scheduled Tasks (Agent workflows)
- OpenAI Realtime audio + Gemini Multimodal support (#4382)
- User-provided API Keys for Tools + RAG API (#4758)
- If
user_providedAPI KEY value is used, fetch models for user (#1028) - Conversation/file retention settings (#2365)
- Conversation/message limits per user (#3058)
- File storage limits per user (#3057)
- Organize chats into folders/projects (#4848)
- Allow Title Endpoint/Models to be used for different Endpoints (#3321)
- Setting to Display the model name in the conversation (#3203)
- Allow import/export function for all chats (#3768)
- S3 and S3-like File Storage support (#2898)
- Send messages that contain only images (#3157)
Partnerships
Besides the awesome contributions from the open-source community, LibreChat will have some help developing this year!
A strategic partnership will enlist developers to contribute the following to open-source this year:
- RAG API Optimizations and Improvements, for higher quality semantic search
- An Agent Marketplace UI, functionally similar to https://chatgpt.com/gpts but for LibreChat Agents
- Better/more granular access controls for Agent Sharing
The core LC team will help lead these improvements but will not be as involved in early/prototype stages, allowing more time for planned items.
There are other partnerships still in the works, so stay tuned!
Beyond the Core Project
Here are some items that are also planned this year, but less involved with the core project.
- LC Code Interpreter API - Pay-as-you-go Pricing, ETA TBD.
- Fully hosted, public LibreChat offering, likely launching Q3.
Proposed Items
Additional items that were not mentioned—whether already proposed or suggested in the future by the community, in both our Discussions and Issues pages—will continue to be reviewed and may be implemented based on demand and contributions. The best way to support one of these items is to leave a reaction on their appropriate page (all of them will be programmatically tallied for a thorough analysis), as well as mention them in the dedicated Roadmap threads linked below.
Conclusion
Thank you all for using and improving LibreChat. I’m very optimistic about 2025 being our biggest year of updates, new features, and improvements.
Please share your thoughts in the dedicated discussion threads, we’ve love to hear feedback and suggestions!