[0.5.1] – 2024-12-25
Added
- 🔕 Notification Sound Toggle: You can now disable notification sounds to help maintain your focus in the workspace. Find this setting under Settings > Interface.
Fixed
- 🔄 Non-Streaming Response Visibility: Resolved an issue that caused non-streaming responses to not display, ensuring that all responses are reliably shown in your conversations.
- 🖋️ Title Generation: Fixed a bug preventing automatic title generation when using OpenAI APIs, restoring smooth organization for your chats.
- 🖼️ Image Generation Error: Fixed an error related to image generation, ensuring a seamless experience in your creative workflows.
- ⚙️ Pipeline Settings Loading Issue: Resolved a problem where pipeline settings were stuck on the loading screen, restoring full access to configuration options.
Enjoy the new enhancements in your user experience!
[0.5.0] – 2024-12-25
✨ Exciting New Features ✨
- 💬 Truly Asynchronous Chat: Start a chat, go do other things, and come back when the AI is ready with a response! Great for complex tasks and multi-agent workflows.
- 🔔 Chat Notifications: Get notified right in the app when your chats finish, even if you’ve switched tabs!
- 🌐 Webhook Notifications: Get chat updates sent to a webhook URL (Settings > Account) even if the app is closed!
- 📚 Channels (Beta): Try out Discord/Slack-like chat rooms for real-time collaboration with other users and AIs. Build bots and unlock multi-agent workflows! Enable in Admin Settings > General.
- 🖼️ Faster Image Uploads: Upload images faster by compressing them before sending (Settings > Interface).
- 🛠️ Group Control with OAuth: Manage your teams easily using OAuth integration.
- ✅ More Accurate Results with Ollama: Send structured output directly to Ollama for more precise data handling.
- 📸 Quick Screen Capture: Take screenshots in one click from the message input menu.
🛠️ Bug Fixes 🛠️
- 📋 Reliable CSV Exports: Fixed issues with missing headers or errors when exporting data with commas.
- 🔓 Access Controls: Fixed an issue preventing some users from accessing models when using ‘BYPASS_MODEL_ACCESS_CONTROL’.
⚠️ Important Changes ⚠️
- ⚙️ Backend Changes: Important backend updates made. Functions that use internal variables may need adjustment. Websocket support is now required.
- 🔑 API Security: API keys are now more secure, only working for model lists and chat completion.
[0.4.8] – 2024-12-07
✨ New Features ✨
- 🔓 Easier Model Access: Easily bypass access controls when you don’t need them using the new ‘BYPASS_MODEL_ACCESS_CONTROL’ variable.
- 📝 Markdown Banners: Make announcements more exciting with markdown in banners.
- 🎨 Polished Look: General UI style refinements.
- 📋 Reliable Rich Text: Improved rich text input in chats.
🛠️ Bug Fixes 🛠️
- 💡 Smoother System: Fixed a breaking bug related to Tailwind, ensuring a more reliable experience.
- 📚 Knowledge Base Fix: Improved the accuracy of information retrieval in the knowledge base.
[0.4.7c] – 2024-12-06
✨ Improvements ✨
- ✨ More Personal Experience: Customize themes, layouts, and content preferences based on your behavior and feedback.
Changed
- 🎨 Smooth Transitions: Fine-tuned animations for a better visual flow.
- 🔧 Consistency: Fixed issues with Tailwind CSS to improve design across all devices.
[0.4.7b] – 2024-12-04
✨ New Features ✨
- ✨ Smooth Animations: Enjoy a more fluid and engaging experience throughout the app.
- 🔌 Extensibility: Enhance the platform with new tools and integrations using the new extension feature.
- 🖥️ Code Readability: Display line numbers in code snippets for better readability.
Changed
- 🎨 Diagram Improvements: Revamped the UI for Mermaid diagrams, making flowcharts more interactive.
[0.4.7a] – 2024-12-01
Changed
- 🎨 Home Page Makeover: The home page has been updated with a more modern look and easier navigation.
[0.4.7] – 2024-12-01
✨ New Features ✨
- ✨ Smart Prompt Input: Let the AI suggest and complete your prompts as you type! Just press “Tab” or swipe right (default with Rich Text Input).
🛠️ Bug Fixes 🛠️
- 🛠️ Tool Exports: Fixed an issue where exporting tools wasn’t working.
- 🔗 Model Setup: Model IDs now register correctly.
- 🖋️ Textarea Improvements: Text areas expand automatically, and Ollama /embed endpoint works as expected.
Changed
- 🎨 Knowledge Base Style: Knowledge base visuals have been cleaned up for a more modern look.
[0.4.6a] – 2024-11-28
Added
- 📜 Terms of Service & 🔒 Privacy Policy: Outlined terms of service and privacy measures.
Fixed
- 🔍 Search Settings & 🔧 Default Model: Improved search settings and set a default model.
- 🎨 Knowledge Prompt: Changed color of the add knowledge tools prompt for improved visibility
[0.4.6] – 2024-11-26
Fixed
- ✏️ Smooth Typing: Resolved an issue where the text area shifted unexpectedly.
- ⚙️ Configuration: Fixed models configuration modal.
- 🔍 Legacy Query Support: Improved web search when using older prompts.
- ⚡ General Reliability: Minor fixes for a smoother experience across the board.
[0.4.5a] – 2024-11-27
Fixed
- 🔧 Configuration Consistency: Ensured the default usage setting is set correctly.
Added
- 📝 Base Model System Prompt: Added a base system prompt for smoother model interactions.
[0.4.5] – 2024-11-26
Added
- 🎨 Model Management: Reintroduced the ability to set model order and default models in Admin Settings > Models.
Fixed
- 🔍 Improved Web Searches: Resolved an issue in web search query generation.
- 📏 Textarea Fix: Fixed the height of the text area when editing system prompts.
- 🔑 Ollama Authorization: Ensured reliable authorization with Ollama’s headers.
- ⚙️ Min_P Parameter: Fixed the ‘min_p’ parameter saving issue.
- 🛠️ Tool Descriptions: Included tool descriptions in the payload.
[0.4.4] – 2024-11-22
Fixed
- 📱 Mobile Navigation: Fixed controls button visibility on mobile devices.
- 📷 LDAP Image: Fixed LDAP integration issues with profile images.
- ⏳ RAG Query Speed: Improved RAG query generation speed when no files are attached.
Changed
- ⚙️ Tool Compatibility: Reintroduced legacy “citation” support to event emitters in tools.
[0.4.3] – 2024-11-21
Added
- 📚 Inline Citations for RAG: Get inline citations for RAG using the default RAG prompt with new files.
- 🎨 Rich Text Improvement: Smoother and more reliable rich text formatting in chats.
- ⚡ Faster Model Loading: Faster model retrieval and better performance.
Fixed
- 🔗 Pipelines Functionality: Resolved issues with pipelines.
- ✏️ Ollama Suffix: Added the missing “suffix” field to the Ollama generate form.
Changed
- 🗂️ Citations to Sources: Renamed “citations” to “sources” for consistency.
[0.4.2] – 2024-11-20
Fixed
- 📁 Knowledge File Visibility: Fixed displaying files when referenced with ‘#’.
- 🔗 OpenAI Endpoint Fix: Fixed issues with non-official OpenAI connections using prefixes.
- ⚔️ Arena Model: Fixed an issue with saving arena model access controls.
- 🔧 Model Usage: Fixed the usage selector in the model editor.
[0.4.1] – 2024-11-19
Added
- 📊 Detailed Feedback System: New 1-10 rating scale to give more precise feedback.
- ℹ️ Tool Descriptions on Hover: Hover over the message input to see tool descriptions, giving more context.
Fixed
- 🗑️ User Deletion: Deleted users now don’t cause errors in workspaces.
- 🔑 API Key Creation: Fixed creating API keys.
- 🔗 HTTPS Proxy Fix: Corrected proxy issues affecting model management.
[0.4.0] – 2024-11-19
✨ Big Update! ✨
- 👥 User Groups: Create and manage user groups for better organization.
- 🔐 Group-Based Access: Set access to models, knowledge, prompts, and tools based on user groups for better control.
- 🛠️ Permissions: Manage user permissions (upload files, edit chats, etc.) on a group level.
- 🔑 LDAP Support: New LDAP authentication for better security.
- 🌐 Enhanced OpenAI Connections: Added prefix ID support to avoid model ID clashes.
- 🔐 Ollama API Key Support: Manage Ollama credentials behind proxies, including prefix ID.
- 🔄 Enable/Disable Connections: Enable/disable OpenAI and Ollama connections as needed.
- 🎨 Redesigned Model/Prompt Workspace: Fresh designs for easier model and prompt management.
- 🧩 Sorted Functions: Functions are now automatically categorized for better management.
- 💻 Redesigned Collaboration: Improved support for multiple users contributing to models, knowledge, prompts, or tools.
- 🔧 Auto-Selected Tools: Tools enabled through the model editor are now automatically selected.
- 🔔 Search & Tool Indicator: Easily see when web search or tools are active.
- 🔑 Toggle API Key Auth: Option to disable API authentication for Blizzer.
- 🗂️ Agentic Retrieval: Better RAG accuracy by analyzing chat history for optimal queries.
- 📁 Large Text to File: Option to convert large pasted text into a file upload.
- 🗂️ Toggle Citations for Models: Added the ability to disable citations.
- 🔍 Settings Search: Easily search for settings fields.
- 🗣️ SpeechT5 TTS: Added experimental local SpeechT5 text-to-speech capabilities.
- 🔄 Unified Reset: One-click option to reset and remove all models from settings.
- 🛠️ Initial Setup Wizard: More clarity during the first-time setup.
Fixed
- 🎥 YouTube Video Attachments: Fixed loading issues with YouTube videos.
- 🔄 Shared Chat Updates: Shared chats now update correctly.
- 🔍 DuckDuckGo Fix: Addressed search integration issues.
- 🧾 Citations: Adjusted relevance calculation for citations, so Blizzer reflects source accuracy in RAG.
- 🔑 Jina API Key: Added an option to enter an API key for Jina Search.
Changed
- 🛠️ Functions in Admin Panel: As functions are advanced plugins, they’re now accessible from the Admin Panel instead of the workspace.
- 🛠️ Manage Ollama Connections: Ollama connections now managed from Admin Settings > Connections.
- 📊 Base Models in Admin: Admins now can see base models (connections or functions), in “Models” Admin settings.
- 📌 Model Selection Persists: Model chosen in a previous chat remains for new chats.
- 🎨 Design Improvements: General design refinements across the platform.
Removed
- 📂 Model Reordering & ⚙️ Default Model: Temporarily removed, will be back in user group settings.
[0.3.35] – 2024-10-26
Added
- 🌐 Translation Update: New translation labels and updated Brazilian Portuguese translation (pt-BR)
- 📁 Clear File Upload Warnings: If file content extraction fails, you’ll now get a warning.
Fixed
- 📚 Knowledge Base: Fixed an issue where the Knowledge Base wasn’t loading.
- 🛠️ Tool Parameters: Tools now function correctly with all parameters.
- 🔗 Multi-Model Chats: Responses in multi-model chats are no longer deleted.
[0.3.34] – 2024-10-26
Added
- 🔧 JSON Feedback Export: Feedback data can now be exported to JSON.
- 🗂️ Smart Search: Optimized search by only loading embedding models when needed.
- 🎨 Toggle Text Input: Switch back to a simpler text input for chat if you prefer.
- 🛠️ Reliable Tools: Improved method for calling tools.
Fixed
- 🖥️ Rename Folders: Fixed folder renaming issues in Firefox.
- 🔠 Text Splitter: Tiktoken text splitter now works correctly in Docker.
- 💼 S3 File Uploads: Fixed S3 file upload issues.
- 🔒 Secure Connection: Resolved a crash related to Strict-Transport-Security (HSTS).
- 🚫 Login Reliability: Addressed issues with boolean values during login.
- ⚙️ Rich Text Paste: Smoother and more intuitive rich text pasting.
- 🔨 Model Filtering: Corrected model filter issue.
- 🏷️ Tag Fix: Fixed an issue preventing custom “tags generation prompts” from registering properly.
[0.3.33] – 2024-10-24
Added
- 🏆 Evaluation Leaderboard: Track your performance with a new real-time leaderboard system based on the Elo rating system.
- ⚔️ Model A/B Testing: Directly compare models side-by-side in a blind A/B test.
- 🎯 Topic-Based Leaderboard: More accurate rankings based on specific topics.
- 📁 Folders for Chats: Organize chats into folders for easy sharing and analysis.
- 📤 Drag & Drop Chat Import: Import chats by simply dragging and dropping them into the sidebar.
- 📚 Enhanced Knowledge Collection: Reference specific files within knowledge collections.
- 🏷️ Improved Tagging: New ‘tag:’ query system for better conversation management.
- 🧠 Auto-Tagging: Your conversations will be automatically tagged for better organization.
- 🔍 Backend Chat Search: Faster and more accurate chat filtering using the backend.
- 🎮 Revamped Playground: Updated playground for testing models and tools.
- 🧩 Token-Based Text Splitter: More control over text processing using token-based splitting.
- 🔢 Ollama Batch Embeddings: Faster performance with Ollama embedding models.
- 🔍 Enhanced Knowledge Input: Cleaner workflow for adding knowledge content.
- 🖋️ Rich Text Input: Format your chats with support for rich text formatting.
- ⚡ Faster Whisper: Customize your local faster whisper model directly in the app.
- ☁️ Experimental S3 Support: Use S3 support for stateless setups.
- 🔕 Disable Update Notifications: Option to turn off update notifications.
- 🌟 RAG Accuracy: Added relevance percentages in RAG results.
- ⚙️ Mermaid Copy: Added a copy button for mermaid diagrams.
- 🎨 UI Redesign: Major interface redesign for a smoother, more modern experience.
Fixed
- 🎙️ Voice Note Fix: Microphone no longer stays active after finishing a voice note recording.
Removed
- 👋 Sidebar Tags: Removed tags from sidebar, now using query-based tag filtering.
[0.3.32g] – 2024-10-24
Added
- 🔑 User Token Usage: Track user token usage.
- 💸 Spending Report: New report to help users track their spending.
- 💵 Updated Model Pricing: Updated model pricing.
[0.3.32f] – 2024-10-21
Added
- 📊 Activity Tracking: Monitor user activity patterns.
- 💰 Pricing Page: Dedicated page outlining subscription plans and features.
[0.3.32e] – 2024-10-17
Added
- 📧 Email Verification: Verify user email addresses.
- 👤 Display Email Name: Display user names with email addresses.
Changed
- 🎨 New Color Scheme: Improved aesthetic and user interface coherence.
Fixed
- 🖼️ Image Loading: Fixed image loading size for better performance.
[0.3.32d] – 2024-10-15
Added
- 📂 Drag and Drop Import: Import chat history via drag and drop.
Fixed
- ⚙️ Settings: Fixed issue with applying default settings.
- 🛠️ New Chat: Fixed issue with the new chat feature.
Changed
- 🎨 Email Style: Improved style for sending emails.
[0.3.32c] – 2024-10-08
Added
- 📧 SMTP Feature: Improved email handling and notifications.
Fixed
- ✏️ Email Editing: Fixed text editing issue in the email modal.
[0.3.32b] – 2024-10-08
Added
- 📧 SMTP Feature: Improved email handling and notifications.
Changed
- 🔄 Chat Button: Updated the styling of the new chat button.
- 🎨 Image Loading: Added loading indicator during image generation.
- ⏳ Mermaid Indicator: Added a loading indicator for Mermaid rendering.
[0.3.32a] – 2024-10-07
Changed
- 🔄 Styling: Changed suggested and sidebar color.
[0.3.32] – 2024-10-06
Added
- 🔢 Workspace Display: Added count for models, prompts, tools, and functions.
Fixed
- 🖥️ Attachments: Fixed attaching web links and YouTube videos.
- 📞 Call Mode: Fixed call mode on the landing page.
Changed
- 🔄 URL Parameters: Updated the ‘tool_ids’ URL parameter for clarity.
- 🎨 Button Styling: Updated floating buttons styling for better visibility.
- 🔧 Enhanced Accessibility: Floating buttons now close with ‘Esc’ key.
- 🖇️ Information URL: Updated the information URL to a general release page.
- 📦 Dependency Update: Upgraded dependencies for pip installs.
[0.3.31] – 2024-10-06
Added
- 📚 Knowledge Feature: Enhanced document management with improved performance, better UI, and easier integration with RAG.
- 🌐 New Landing Page: New landing page design with an option to switch to the classic chat UI in settings.
- 📁 Full Document Retrieval: Switch between full document retrieval or traditional snippets.
- 📄 Extracted Content Display: View extracted content directly by clicking on the file item.
- 🎨 Artifacts Feature: Render web content and SVGs directly in the app.
- 🖊️ Editable Code Blocks: Live edit code blocks in the LLM response.
- 🔧 Code Block Enhancements: Copy code blocks more easily using the new button.
- 🔍 SVG Pan/Zoom: Enhance interaction with SVG images with pan and zoom.
- 🔍 Text Selection Actions: Perform actions like “Ask a Question” directly on text selections.
- 🗃️ Database Pool: Enhanced database handling for scalability.
- 🔊 Experimental Audio Compression: Compress audio files for speech-to-text.
- 🔍 Query Embedding: Reduced repeated query embedding for better performance.
- 💾 Lazy Loading: Improved load times with lazy loading of large dependencies.
- 🍏 Apple Touch Icon: Optimized display of icons on Apple mobile devices.
- 🔽 Markdown Content: Added support for expandable markdown content.
Fixed
- 🔘 Action Button: Fixed an issue with action buttons.
- 🔄 Multi-Model Loop: Fixed an infinite loop issue in multi-model chats.
- 📄 Chat Exports: Resolved issues with exporting chats to PDF and TXT formats.
- 🔊 Text-to-Speech: Rectified issues with text-to-speech.
Changed
- ⚙️ Endpoint Renaming: Renamed ‘rag’ endpoints to ‘retrieval’.
- 🎨 UI Updates: Refinements across the platform to improve visual appeal.
Removed
- 🗑️ Deprecated ‘DOCS_DIR’: Removed outdated ‘docs_dir’ variable in favor of file management solutions.
[0.3.30g] – 2024-09-27
Changed
- PWA Icon: Changed size of Progressive Web App (PWA) icon.
[0.3.30f] – 2024-09-27
Fixed
- Docker Production Build: Eliminated inclusion of ‘src’ for better efficiency.
- PWA Icon: Resolved issue with the Progressive Web App (PWA) icon.
[0.3.30e] – 2024-09-27
Fixed
- Docker Build: Removed src folder from Prod build.
[0.3.30d] – 2024-09-27
Changed
- Docker Build: Docker build with NODE_ENV=production.
[0.3.30c] – 2024-09-26
Changed
- 🍞 Suggested Prompt: The suggested prompt is now not random.
[0.3.30b] – 2024-09-26
Fixed
- 🍞 Chat Scrolling: Smoother chat scrolling animations.
- 🔑 Keyboard: Fixed issue with the keyboard staying open when the sidebar is active.
[0.3.30] – 2024-09-26
Fixed
- 🍞 Update Available: Dismissed update notification won’t reappear for 24 hours.
- 📋 Ollama Data: Integration of /embed form data now matches Ollama’s specs.
- 🔧 Max Tokens: Fixed compatibility issue with OpenAI ‘o1’ models max_completion_tokens.
- 🔄 Pip Install: Fixed database issues during pip installs.
- 🏷️ Chat Renaming: Browser tab title changes when renaming a chat.
[0.3.29] – 2023-09-25
Fixed
- 🔧 KaTeX Rendering: Enhanced display of complex mathematical notation.
- 📞 Voice Call Activation: Corrected voice call activation through URL triggers.
- 🔄 Configuration Reset: Settings now revert to default correctly upon startup.
- 🌍 Filter Functionality: Ensured all filter functions work as intended.
[0.3.28] – 2024-09-24
Fixed
- 🔍 Web Search: Fixed issue with web search not functioning.
[0.3.27] – 2024-09-24
Fixed
- 🔄 Periodic Cleanup: Resolved runtime error related to cleanup, ensuring smoother performance.
- 📊 Enhanced LaTeX: Improved LaTeX rendering.
[0.3.26] – 2024-09-24
Fixed
- 🔄 Event Loop: Fixed issue causing ‘periodic_usage_pool_cleanup’ to fail with pip installs.
[0.3.25] – 2024-09-24
Fixed
- 🖼️ Image Generation: Fixed issue with image generation.
- ⚖️ Rating Responses: Fixed rate response mechanism.
[0.3.24] – 2024-09-24
Added
- 🚀 Performance Optimization: Improved message rendering performance.
- 💖 Favorite Responses: Mark responses as favorite in the chat overview.
- 💬 New Message Shortcut: Create new message pairs with Cmd/Ctrl+Shift+Enter.
- 🎵 Enhanced Audio: Now supports ‘audio/x-m4a’ files.
- 🔏 Model Selection: Select model using URL parameters.
- 📄 PDF Citation: PDF citations now open to the specific page.
- 🔧 Redis Sockets: Enhanced socket implementation to fully support Redis.
- 🌍 Model Streaming: Stream individual model responses with custom settings.
- 🕒 Model Details: Show model hash and last modified timestamp for Ollama.
- ❗ Admin Notification: Admins receive immediate update info upon login.
Fixed
- 🗑️ File Handling: Fixed errors on Windows related to accessing temporary files.
- 🔓 Authentication: Corrected issue with disabling authentication via ‘WEBUI_AUTH=False’.
- 🔧 Save as Copy: Fixed issue where you could not save messages as copies in many model chats.
- 🔒 Sidebar Closure: Sidebar now closes on mobile after engaging with the menu.
- 🛡️ Tooltips Security: Resolved a security vulnerability in tooltips.
Changed
- ↩️ Stream Response: Moved interface stream response settings to advanced parameters.
- ⚙️ Renamed ‘speedRate’ to ‘playbackRate’: For clarity in media settings.
[0.3.23] – 2024-09-21
Added
- 🚀 WebSocket Redis: Improved load balancing for multiple instances.
- 🔧 Chat Controls: Width-adjustable chat controls for better customization.
- 🌎 i18n Updates: Improved Chinese translations.
Fixed
- 🌐 Task Model Loading: Fixed issue of task models unloading.
- 📝 Title Generation: Fixed title generation for OpenAI compatible APIs.
- 🗂️ RAG Duplicates: Fixed duplicate processing of uploaded files.
- 🖼️ Image Generation: Image generation is now asynchronous and won’t freeze the UI.
- 🔓 Downgrade Authlib: Reverted to fix OAuth issues.
Changed
- 🔍 Message Interaction: Easier focus redirection with a simple click.
- ✨ Styling: Cleaner, more modern styling throughout the UI.
[0.3.22] – 2024-09-19
Added
- ⭐ Chat Overview: Node-based interactive diagram for visualizing conversations.
- 🔗 Multiple Vector DB Support: Added support for multiple vector databases, including Milvus.
- 📡 Experimental Non-Stream: Experimental feature for OpenAI ‘o1’ models.
- 🔍 Experimental Reranking: Added support for “jinaai/jina-colbert-v2” as a reranker.
- 🕸️ Websocket Ignore: Option to ignore websocket upgrades on platforms with issues.
- 🔊 Azure Speech: Added support for Azure Speech services for Text-to-Speech (TTS).
- 🎚️ Playback Speed: Customizable playback speed in Call mode.
- 🧠 Error Messaging: Displays helpful error messages during chat completion issues.
- 📂 Transparent PNG: Model profile images are now saved as PNGs.
- 📱 iPhone Layout: Added padding for iPhone navigation bar.
- 🔗 Secure Headers: Implemented security response headers.
- 🔧 Enhanced ComfyUI: Configure ‘CFG Scale’, ‘Sampler’, and ‘Scheduler’ parameters directly in the admin settings.
- 🌍 i18n Updates: Improved translations for Chinese, Ukrainian, Russian, and French.
Fixed
- 🛠️ Message Deletion: Resolved issues with chat message deletion.
- 🔢 Ordered List: Fixed incorrect list numbering.
Changed
- 🎨 Transparent Icons: Model icons can now be displayed on transparent backgrounds.
- 📝 RAG Template: Enhanced RAG template for better error handling.
[0.3.21] – 2024-09-08
Added
- 📊 Document Count: Total number of documents is now displayed.
- 🚀 Ollama Embed: Added /api/embed endpoint proxy support.
Fixed
- 🐳 Docker Launch: Fixed issue preventing proper launch when using Docker.
Changed
- 🔍 Enhanced Search Prompts: Improved search query generation for better accuracy.
[0.3.20] – 2024-09-07
Added
- 🌐 Translation Update: Updated Catalan translations.
Fixed
- 📄 PDF Download: Resolved issue to download PDFs with correct formatting.
- 🛠️ Tool Requirements: Fixed install of tool requirements.
- 🔗 Inline Image Links: Enabled rendering of images directly from links.
- 📞 UI Cleanup: UI automatically closes after voice calls.
- 🎙️ Post-Call Deactivation: Microphone now deactivates after calls.
- ✍️ Markdown Spacing: Corrected spacing in Markdown rendering.
- 🔄 Message Re-rendering: Fixed re-rendering of response messages.
Changed
- 🌐 Web Search: Removed the prompt threshold and added a toggle button to enable Web Search query generation.
- 📝 Default Templates: Emptied environment variable templates now default to Blizzer defaults.
[0.3.19] – 2024-09-05
Added
- 🌐 Translation Update: Improved Chinese translations.
Fixed
- 📂 DATA_DIR: Fixed overriding DATA_DIR issues.
- 🛠️ Frontmatter Extraction: Fixed frontmatter extraction for tools and functions.
Changed
- 🎨 UI Styling: Refined styling for better user experience.
[0.3.18] – 2024-09-04
Added
- 🛠️ Tools & Functions: Python files for tools and functions now load directly from the database.
Fixed
- 🔄 Imports: Tool and function imports automatically renamed to ‘open_webui.’.
- 🎨 Styling: Minor style fixes.
[0.3.17] – 2024-09-04
Added
- 🔄 Import/Export Config: Import and export webui configurations.
- 🌍 Web Search: Activate web search directly via URL: ‘web-search=true’.
- 🌐 SearchApi: Added support for SearchApi web search.
- 🔍 Literal Tool Type: Tools now support the Literal type.
- 🌍 Updated Translations: Improved translations for Chinese, Ukrainian, and Catalan.
Fixed
- 🔧 Pip Install: Resolved issue with pip install failure due to missing ‘alembic.ini’.
- 🌃 Theme Update: Fixed color theme not updating with system changes.
- 🛠️ User Agent in ComfyUI: Added default headers in ComfyUI to fix access issues.
- 🔄 Response Headers: Ensured proper return of proxied response headers.
- 🔗 WebSocket: Modified socket.io to prefer websockets.
- 🎭 Accessibility: Added ARIA labels for buttons.
- ⚖️ Advanced Parameters: Advanced parameters are applied correctly.
Changed
- 🔁 Namespace: Reorganized Python files to ‘open_webui.’ namespace.
- 🚧 Dependency Updates: Updated various backend dependencies for performance.
[0.3.16] – 2024-08-27
Added
- 🚀 Config Migration: Configuration handling now happens through the database, enabling high-availability setups.
- 🔗 Call Mode Shortcut: Activate call mode directly from URL: ‘call=true’.
- ✨ TTS Control: Control how message content is segmented for TTS.
- 😄 Knowledge Status: Show status when using knowledge-augmented models.
- 👆 Code Span Copy: Click to copy code spans directly.
- 🚫 Model Filter for API: Block API users based on model filters.
- 🎬 Call Overlay: Call overlay now only covers the chat control area.
Fixed
- 🔧 LaTeX: Fixed rendering of LaTeX.
- 📁 File Leak: Prevented access to uploaded files across user chats.
- 🔧 Pipe Functions: Fixed the files param in pipe functions.
- 📝 Markdown in RAG: Fixed issues processing Markdown in files.
- 🚫 Duplicate Prompts: Fixed system prompt duplication.
Changed
- 🔋 Wakelock: Only activates during call mode.
- 🔍 Ollama Content: Added ‘application/x-ndjson’ to Ollama responses.
- ✋ Disable Signups: Disabled sign-ups when ‘ENABLE_LOGIN_FORM’ is false.
[0.3.15] – 2024-08-21
Added
- 🔗 Temporary Chat: Enable temporary chats from the URL: ‘temporary-chat=true’.
- 🌄 ComfyUI Seed Node: Set seed IDs for randomized seed assignments in ComfyUI.
Fixed
- 🛠️ Tools and Functions: Resolved issues with tools and functions.
- 🔘 Chat Buttons: Fixed chat action buttons in many model chat environments.
- ⏪ Many Model Compatibility: Restored compatibility for many model chats.
[0.3.14] – 2024-08-21
Added
- 🛠️ Custom ComfyUI Workflow: New customizable workflow.
- 🔀 Merge Responses: Merged responses from multiple models for better interaction in many model chats.
- ✅ Same Model Multiple Times: Add multiple instances of the same model to a chat.
- 🔧 Model Workspace Actions: Quick actions to hide/unhide or delete models.
- 🗨️ Markdown User Messages: User messages are now rendered in Markdown.
- 💬 Temporary Chats: New temporary chat feature.
- 🖋️ User Message Editing: Edit messages without sending.
- 🛡️ Security: Various security improvements.
- 🌍 Updated Translations: Improved translations for Chinese, Ukrainian, and Bahasa Malaysia.
Fixed
- 📑 Mermaid Rendering: Fixed issues with Mermaid charts.
- 🎭 PWA Icon: Fixed the PWA icon to be maskable.
- 🔀 Cloned Chat Freezing: Fixed issues with cloning multi-model chats.
- 🔍 Error Handling: Minor fixes for a smoother experience.
Changed
- 🖼️ Image Generation: Overhauled image generation.
- 🔨 Tool and Function Calling: Refactored tool and function calls for clarity.
- 🌐 Backend Library: Updated backend libraries for better performance.
Removed
- 🚫 ComfyUI Env Vars: Removed outdated ComfyUI variables.
[0.3.13] – 2024-08-14
Added
- 🎨 Improved Markdown: Enhanced rendering for LaTeX and Mermaid charts.
- 🔄 Auto-Install Dependencies: Blizzer now automatically installs extra Python requirements for ‘Tools’ and ‘Functions’.
- 🌀 OAuth Email Customization: Customize the default “email” claim in OAuth using a new variable.
- 📶 Websocket Reconnect: Automatically reconnect when a websocket is closed.
- 🤳 Haptic Feedback: Android devices now support haptic feedback.
Fixed
- 🛠️ ComfyUI Performance: ComfyUI image generation now runs in a separate thread to prevent UI freezing.
- 🔀 Session Handling: Fixed issue that required session_id on the client-side.
- 🖋️ Minor Bug Fixes: Corrected typos, formatting and tests.
Changed
- 🚀 SvelteKit 2: Updated to SvelteKit version 2 for better performance.
- 🧹 General Cleanup: Code cleanup and refactoring for better efficiency.
- 🚧 Integration Testing: Improved Cypress integration tests for better reliability.
- 📁 Standardized File: Renamed file extension to ‘.safetensors’.
Removed
- 🗑️ Frontend Functions: Removed frontend functions migrated to the backend.
[0.3.12] – 2024-08-07
Added
- 🔄 Infinite Scroll: Added infinite scroll to the sidebar for efficient chat navigation.
- 🚀 Markdown Rendering: Improved code blocks, clickable images, and codespan styling.
- 🔒 Admin Access: Admins no longer have visibility over shared chats by default.
Fixed
- 📊 Markdown: Resolved markdown rendering issues.
- 🛠️ Styling: Multiple fixes applied to styling.
- 🗃️ Modal Handling: Fixed modals not closing in some model chat scenarios.
- 📄 OpenAI Usage: Fixed display of usage statistics for OpenAI.
- 🔧 Functions Plugin: Fixed issue with non-streaming operations in Functions.
- 🔄 Environment Variable: Corrected the data type for COMFYUI_FLUX_FP8_CLIP.
Changed
- 🔧 Backend Dependencies: Updated backend dependencies for security and performance.
[0.3.11] – 2024-08-02
Added
- 📊 Model Selection: Added images next to model names for easier model selection.
- 🗣 ElevenLabs: Support for ElevenLabs voice ID by name.
- ⌨️ Model Selection: Easier model selection using arrow keys.
- 🔍 Fuzzy Search: Locate models with fuzzy search.
- 🕹️ ComfyUI Flux: Support for the Flux image gen model and settings.
- 💾 File Sizes: File size displayed for uploads.
- 🎚️ Advanced Params: Added “Min P” parameter for model control.
- 🔒 OAuth: Custom redirect URI support for OAuth behind reverse proxies.
- 🖥 Enhanced Latex Rendering: Adjustments for accurate latex detection and presentation.
- 🌐 Internationalization: Improved translations in Romanian, Vietnamese, and Ukrainian.
Fixed
- 🔧 Tags: Tags are now properly sent on document upload.
- 🖥️ Input Fields: Fixed issues with secure input fields.
- 📂 PDF Paths: Fixed static paths for PDF generation.
Changed
- 🎨 UI/UX: Multiple UI and UX updates for cleaner experience.
- 🚧 Component Refactoring: Refactoring across components.
- 🎛️ User Valves: Moved user valves to direct chat controls.
Removed
- ⚙️ Health Check Logging: Reduced logging for backend performance.
[0.3.10] – 2024-07-17
Fixed
- 🔄 Improved File Upload: Added animation to file uploads.
- 💬 Chat Continuity: Fixed issues with existing chats.
- 🗂️ Chat File Reset: Chat files now reset for new conversations.
- 📁 Workspace Uploads: Corrected document uploads in the workspace.
[0.3.9] – 2024-07-17
Added
- 📁 Files Chat Controls: Manage files directly within chat controls.
- 🔧 “Action” Function: New function for custom buttons on message toolbars.
- 📜 Citations: Citations display the filename and can be opened in a new tab.
- 🛠️ Event Call Updates: Enhanced ‘event_emitter’ for message replacement and ‘event_call’ to support text input.
- 🎨 Styling Refactor: Various styling updates for a cleaner user interface.
Fixed
- 🔧 Chat Controls Priority: Chat controls are now prioritized over model info.
- 🪲 Debug Logs: Fixed logging issues.
- 🔑 Automatic1111: Auth key is no longer required.
- 📝 Title Generation: Title generation runs once, even with multiple models.
- ✅ Boolean Values: Added support for boolean values.
- 🖼️ Files Overlay: Fixed styling of the files overlay.
Changed
- ⬆️ Dependency Updates: Upgraded ‘pydantic’, ‘sqlalchemy’, ‘unstructured’, and ‘chromadb’.
[0.3.8] – 2024-07-09
Added
- 💬 Chat Controls: Adjust chat parameters directly for each session.
- 📌 Pinned Chats: Keep important chats easily accessible.
- 📄 Tika Integration: Apache Tika for document loading.
- 🛠️ Custom OIDC Claims: Set custom claims for OpenID.
- 🔧 Tools & Functions API: Added ‘event_emitter’ and ‘event_call’ with citations.
- ↔️ Settings Scrolling: Settings tabs now support horizontal scrolling.
- 🌑 Darker OLED: New dark theme and improved styling.
Fixed
- ⏰ OpenAI Timeout: Resolved OpenAI streaming timeout with ‘AIOHTTP_CLIENT_TIMEOUT’.
- 💡 User Valves: Fixed malfunctioning user valves.
- 🔄 Collapsible Components: Fixed collapsible components.
Changed
- 🗃️ Database: Switched to SQLAlchemy for better concurrency.
- ⬆️ ChromaDB: Updated to version 0.5.3.
- 🔤 Primary Font: Updated primary font to Archivo.
- 🔄 Font for Windows: Replaced Arimo with Inter font for Windows.
- 🚀 Lazy Loading: Improved performance with lazy loading.
- 📋 Task Generation: Task generations now include only the “task” field.
[0.3.7] – 2024-06-29
Fixed
- 🕵️♂️ Language Detection: Correctly detects and adapts to browser language.
- 🔐 OIDC Role: Admin role is assigned to first user with OIDC.
- 💬 Chat/Completion Endpoint: Endpoint now works correctly with stream set to False.
- 🚫 ‘WEBUI_AUTH’: Correctly applies when set to False.
Changed
- 📦 Dependency: Updated ‘authlib’ for security and performance enhancements.
[0.3.6] – 2024-06-27
Added
- ✨ “Functions” Feature: Use filter and pipe functions directly in the WebUI.
- 📁 Files API: Custom RAG in conjunction with the Filter Function.
- 🛠️ Tool Enhancements: Tools now support citations and “Valves”.
- 🔗 Iframe Support: Render HTML in the chat interface using functions and tools.
- 🔒 Experimental OAuth: New experimental OAuth support.
- 🖼️ Custom Background: Set custom background in Settings > Interface.
- 🔑 AUTOMATIC1111 Auth: Enhanced security for the AUTOMATIC1111 API.
- 🎨 Code Highlight: Improved code highlighting features.
- 🎙️ Voice Interrupt: Toggle voice interruption from Settings.
- 💤 Wakelock API: Wakelock API in use to prevent screen dimming.
- 🔐 API Keys: All API keys are hidden by default.
- 🔍 New Web Search: Added jina_search.
Fixed
- 🔧 Conversation Mode: Fixed conversation mode.
- 📏 Scroll Button: Fixed scroll button obstruction.
Changed
- ⏲️ AIOHTTP_CLIENT_TIMEOUT: Set to ‘None’ by default for better flexibility.
- 📞 Voice Call: Skips code blocks and expressions during calls.
- 🚫 Error Handling: Disabled continuation with error messages.
- 🗂️ Playground: Moved the Playground from the workspace to the user menu.
[0.3.5] – 2024-06-16
Added
- 📞 Enhanced Voice Call: Real-time text-to-speech for each sentence.
- 👆 Tap to Interrupt: Tap to stop speaking in voice calls.
- 😊 Emoji Call: LLMs use emojis in voice calls (toggle from Settings).
- 🖱️ Quick Archive/Delete: Swiftly archive or delete chats using Shift key + mouseover.
- 📝 Markdown Model Descriptions: Format model descriptions with markdown.
- 🧠 Editable Memories: Modify personal information you want the LLM to remember.
- 📋 Admin Sorting: Sort users/chats in the admin panel.
- 🌑 Dark Mode Selectors: Dark mode for chat quick selectors.
- 🔧 Advanced Parameters: Added ‘num_keep’ and ‘num_batch’ to advanced parameters.
- 📅 Dynamic Prompts: Added ‘{{CURRENT_DATETIME}}’, ‘{{CURRENT_TIME}}’, ‘{{USER_LOCATION}}’ variables.
- 🌐 Tavily Search: Includes Tavily as a web search provider option.
- 🖊️ Federated Usernames: Set usernames for federated authentication.
- 🔗 Auto Clean URLs: URLs now have trailing slashes removed automatically.
- 🌐 Enhanced Translations: Improved Chinese and Swedish translations.
Fixed
- ⏳ AIOHTTP_CLIENT_TIMEOUT: New env variable for requests to Ollama longer than 5 minutes.
- ❌ Message Delete Freeze: Fixed issue where message deletion would sometimes cause a freeze.
[0.3.4] – 2024-06-12
Fixed
- 🔒 HTTPS Mixed Content: Fixed issue causing security warnings with mixed content.
- 🔍 Web Search: Reintroduced ‘ENABLE_RAG_LOCAL_WEB_FETCH’ to fix web search functionality.
- 💾 RAG Templates: RAG templates now save correctly.
[0.3.3] – 2024-06-12
Added
- 🛠️ Native Python Functions: Create and integrate custom code within the ‘Tools’ workspace.
- 🌐 DuckDuckGo: Added DuckDuckGo as a web search provider.
- 🌏 Enhanced Translations: Improved Vietnamese and Chinese translations.
Fixed
- 🔗 Web Search URLs: Gracefully handles errors for continuous data loading.
- 🖥️ Frontend Stability: Fixed issues where the frontend stops responding with backend errors.
- 🔧 Pip Install: Fixed issues related to pip installations.
[0.3.2] – 2024-06-10
Added
- 🔍 Web Search Query Status: Web search query persists in the results for easier debugging.
- 🌐 New Web Search Provider: Added Serply as a new option.
- 🌏 Improved Translations: Enhanced translations for Chinese and Portuguese.
Fixed
- 🎤 Audio Files: Fixed bug preventing audio files from being uploaded.
- 💬 Message Input: Improved handling of message inputs and visual feedback.
- ⚙️ Parameter Registration: Fixed issue where parameters weren’t registering properly.
[0.3.1] – 2024-06-09
Fixed
- 💬 Chat: Resolved issue where chat wasn’t working for specific models.
[0.3.0] – 2024-06-09
Added
- 📚 Knowledge for Models: Attach documents directly to models.
- 🎙️ Hands-Free Voice Calls: Initiate voice calls without using your hands.
- 📹 Video Calls: Enable video calls with supported vision models.
- 🎛️ Enhanced Voice UI: Improved user interface for voice recording.
- 🌐 External STT: Support for external Speech-To-Text services.
- ⚙️ Unified Settings: Consolidated all settings into a new admin settings section.
- 🌑 Dark Mode Splash: New splash screen for dark mode users.
- 📥 Upload Pipelines: Upload pipelines directly from settings.
Fixed
- 🛠️ Playground: Fixed issues where playground wasn’t working.
- 🔥 Temperature: Corrected issue with temperature value ‘0’.
- 📝 Input Clearing: Fixed prompt input not clearing right away.
- ✨ Styling: Fixed various UI styling issues.
- 👥 Active Users: Fixed active user display.
- 🌐 Community Platform: Community Platform is back online and compatible.
Changed
- 📝 RAG: Updated RAG to use a system prompt for context.
- 🔄 Settings: Moved Models, Connections, Audio, and Images to admin settings.
- ✍️ Title Generation: Enhanced default prompt for title generation.
- 🔧 Backend Task Management: Tasks now managed on the backend by admin.
- 🔍 Editable Search Query: Edit search query prompts.
- 📏 Prompt Length Threshold: Set prompt length for search generation from admin settings.
- 📣 Settings Consolidation: Merged the Banners with the Interface setting.
[0.2.5] – 2024-06-05
Added
- 👥 Active Users: See how many users are active.
- 🗂️ Create Ollama Modelfile: Reintroduced model file creation.
- ⚙️ Default Model: Set default model from Settings > Interface.
Fixed
- 📱 Mobile View: Improved mobile experience using dvh.
[0.2.4] – 2024-06-03
Added
- 👤 Admin Details: Account pending page displays admin details.
- 🌐 HTTP Proxy: Use ‘http_proxy’ in API calls.
- ❓ Documentation Access: Quick access to documentation on larger screens.
- 🌍 Enhanced Translation: Improvements have been made to translations.
Fixed
- 🔍 SearxNG Search: Fixed the SearxNG web search.
[0.2.3] – 2024-06-03
Added
- 📁 Export Chat: Export chats as JSON from the navbar menu.
- ✏️ Edit Titles: Rename titles with double-click.
- 🧩 Batch Embeddings: Process multiple embeddings in a batch for faster speed.
Fixed
- 🛠️ Modelfile Migration: Fixed modelfile migration script.
- 💬 Zhuyin Input: Fixed issue with Zhuyin input on Mac.
- 🔊 Local TTS Voice: Fixed display of the selected local TTS voice.
[0.2.2] – 2024-06-02
Added
- 🌊 Mermaid Support: Create diagrams and flowcharts directly within Blizzer.
- 🔄 Config Reset: Reset configuration settings on start using ‘RESET_CONFIG_ON_START’ variable.
Fixed
- 🔧 Pipeline Filters: Fixed issues with pipeline filters.
[0.2.1] – 2024-06-02
Added
- 🖱️ Single Model Export: Easily export models with one click.
- 🖥️ Ollama Parameters: Added ‘num_thread’, ‘use_mmap’, and ‘use_mlock’ parameters for Ollama.
Fixed
- 🔧 OpenAI URL: Corrected issues with saving OpenAI URL API settings.
- 🚫 Model Display: Fixed displaying models when Ollama is disabled.
Changed
- 💡 Versioning Update: Version 0.2.y will focus on bug fixes, while major updates will be 0.x.
[0.2.0] – 2024-06-01
✨ Big Update! ✨
- 🔧 Pipelines Support: New plugin framework for enhanced customization (https://github.com/open-webui/pipelines).
- 🔗 Function Calling: Integrate function calls seamlessly through Pipelines.
- ⚖️ User Rate Limiting: Implement user-specific rate limits using Pipelines.
- 📊 Langfuse Monitoring: Track usage statistics with Langfuse through Pipelines.
- 🕒 Conversation Limits: Set limits on conversation turns through Pipelines.
- 🛡️ Toxic Filtering: Filter out toxic messages automatically using Pipelines.
- 🔍 Web Search: Built-in web search via RAG using various providers.
- 🗂️ Models Workspace: Create and manage model presets.
- 🛠️ Model Builder: Persistent model builder mode.
- 🏷️ Model Tagging: Organize models with tagging.
- 📋 Model Ordering: Organize models with drag and drop.
- 📈 OpenAI Stats: Detailed stats for OpenAI models.
- 📅 System Prompt Variables: New ‘{{CURRENT_DATE}}’ and ‘{{USER_NAME}}’ variables.
- 📢 Global Banners: Manage global banners from admin settings.
- 🗃️ Archived Chats: Easily search and export archived chats.
- 📂 Archive All: Quickly archive all chats from settings.
Fixed
- 🔍 Archived Chats: Fixed visibility of archived chats in the admin panel.
- 💬 Message Styling: Resolved issues with message appearance.
- 🔗 Shared Chat Responses: Shared chat responses are now read-only.
- 🖥️ UI Enhancement: Fixed scrollbar overlapping issue.
Changed
- 💾 User Settings: User settings now saved on the backend.
- 📡 Unified API Requests: Unified API for getting models ‘/api/models’.
- 🔄 Versioning: Versioning will be 0.x for major updates and 0.x.y for patches.
- 📦 Export All Chats: Moved “Export All Chats” to the Admin Panel settings.
Removed
- 🚫 Bundled LiteLLM: Migrate to self-hosted LiteLLM.
[0.1.125] – 2024-05-19
Added
- 🔄 Updated UI: Chat interface now features chat bubbles (old style is also available via settings).
- 📂 Enhanced Sidebar: Model files, documents, prompts, and the playground are merged into the Workspace.
- 🚀 Improved Multi-Model: All responses now displayed simultaneously.
- 🐍 Python Execution: Execute Python code in the browser.
- 🧠 Experimental Memory: Personal information can now be stored.
- 💾 Persistent Settings: Settings now saved as config.json.
- 🩺 Health Check: Added a health check endpoint.
- ↕️ RTL Support: Toggle chat direction.
- 🖥️ PowerPoint Support: RAG supports PowerPoint documents.
Changed
- 👤 Shared Chat: Shared chat now includes creator info.