Changelog
Follow up on the latest improvements and updates.
RSS
new
fixed
improved
v2.7.0: Welcome o1 models + improved reliability
Thinkbuddy 2.7.0 introduces advanced vision models, new reasoning models, important bug fixes about LLM reliability, smarter file extraction, and significant performance upgrades to supercharge your AI-assisted workflow.
New Features
Enhanced Vision Model Capabilities
- Full Chat Visibility: Vision model now sees entire chat history, improving context awareness.
- Claude Vision Integration: Added support for Claude Vision model.
- New AI Models: Introduced o1-preview, o1-mini, Mixtral, and Llama 3.2.
Advanced OCR System for File Extraction
- Flexible Input: Upload PDFs directly or provide a URL.
- AI-Powered Extraction: Uses GPT-4 Turbo with Vision for accurate text and image extraction.
- Structured Output: Extracted content formatted in Markdown for readability.
Light Theme (Beta)
- New option for users preferring a lighter interface.
Improvements
Settings UI Overhaul
- Redesigned buttons and icons for better user experience.
- Overall aesthetic and usability enhancements.
Performance Optimizations
- Significant CPU efficiency improvements.
- Faster PDF processing with parallel page conversion.
- Enhanced throughput with batch image processing.
- Improved scalability with asynchronous processing.
- Added retry mechanism for handling failures and API limits.
Robust Error Handling
- Comprehensive logging and exception handling for increased reliability.
LLM Model Health
- General stability and performance improvements to various LLM models.
Bug Fixes
- Addressed various issues to enhance overall app stability and performance.
The latest ThinkBuddy update featuring improved selection functionality, optimised AI streaming experience, a new hide hotkey, faster file attachment parsing and important bug fixes.
Improvements
- Enhanced Selection Functionality: The experience of selecting items within the application has been improved for better usability.
- Improved AI Streaming Experience: The performance and responsiveness of the AI streaming feature have been optimized for a smoother experience.
- New Hide Hotkey: Added the shortcut Cmd + Hto quickly hide the application.
- Faster File Attachment Parsing: The speed of parsing file attachments has been significantly improved, resulting in quicker processing times.
Bug Fixes
- Memory Leak Resolved: A memory leak issue has been fixed, enhancing the overall stability and performance of the application.
Discover the latest Thinkbuddy update, featuring enhanced file support across multiple formats, new LLM model integrations, crucial bug fixes, and UI improvements. Check out the details below!
New Features
- Attachment Button Added to Chat Screen
- File System Support for Various Formats (.pptx, .xlsx, .xls, .csv, .docx, .pdf, .xml, .md, .json, .sql, .tex, .wav, .m4a)
- New LLM Models Integrated (Meta Llama 3.1 (405b), GPT-4o Mini, GPT-4o Vision)
Bug Fixes
- Fixed the duplicated launch bug that occurred at startup.
- Resolved the fullscreen bug.
Improvements
- Updated the Message Menu UI.
new
fixed
v2.5.6: Multi-Model & Remix
The Remix feature, which prepares the best results with information from multiple models, has been released.
New Features
- Introduced a new multi-modal AI system.
- Added Gemini Flash feature.
- Integrated Claude 3.5 Sonnet.
- Implemented Remix modal.
- Developed a new instructions system.
- Added a chat recovery system.
- Introduced a new model selection dialog.
- Developed a new inline writer system.
- Implemented a new local storage management system.
Bug Fixes
- Fixed the bug related to editing answers.
- # Improvements
- Reduced latency time.
- Enabled inline writer functionality across all browsers and applications.
- Made performance optimizations.
- Enhanced markdown preview optimizations.
In this update, we've introduced new AI model GPT-4o which is 2x faster then GPT-4 Turbo.
New Feature
- New AI Model Implemented: Now featuring the newest GPT-4o model.
In this update, we've resolved a UI issue affecting some new users and integrated a new inline writer system.
Improvements
- Implemented a new inline writer system to improve user experience. The operating logic of this feature has been updated for increased stability. This feature is currently in testing stage, and a tutorial video will be available in the following days.
- Performance enhancements for smoother operation.
Bug Fixes
- Resolved an UI issue affecting some new users.
In this update, we've introduced new AI models LLaMA3 70b, Gemini 1.5 Pro and new features like markdown preview, voice recording, and chat history export. We've also made improvements to text selection, CPU usage, and UI elements. Additionally, we've fixed bugs related to chat thread continuation, cursor visibility, iCloud sync, and shortcuts.
New Features
- New AI Model Implemented: Now featuring the LLaMA3 70b and Gemini 1.5 Pro ((1 million context length) models.
- Markdown Preview for User Messages: Users can now see a preview of their messages formatted in Markdown.
- Voice Recording for Chat Queries: Initiate voice recordings in chat or quick chat with a convenient voice shortcut.
- Pin Special Prompts: Users have the ability to pin their favorite prompts for quicker access.
- Export Chat History: Chat history can now be exported in JSON format.
- Message Management: New options to delete selected messages, edit messages, and resend edited messages have been added.
- App Position and Window Size: The app now automatically saves its position and window size to local storage.
- Selectable Clipboard Content: Users can now select content directly from the clipboard.
Improvements
- Enhanced Text Selection: The experience of selecting text has been improved for ease of use.
- Text Position Selection: Users can now select text based on its position.
- Local Storage Directory Update: The directory for local storage has been changed to improve organization.
- Cursor Focus: The experience of cursor focus has been updated for better usability.
- CPU Usage Optimization: Background CPU usage has been optimized for improved performance.
- Dialog and Menu Blur Effects: The background blur effect for the AI dialog and prompt menu has been updated for a sleeker look.
- Markdown Code Block UI: Improvements have been made to the UI for markdown code blocks for better readability.
- Recording Voice: Users can now record their voice when assistant writing answers.
Bug Fixes
- Chat Thread Continuation: A bug that affected the saving of chat thread continuations has been fixed.
- Hidden Cursor Issue: The bug causing the cursor to hide has been resolved.
- iCloud Sync: An issue with iCloud sync has been fixed.
- Shortcut Bugs: The bug where the shortcut for adding a new question in quick chat wasn't working has been fixed. Similarly, the issue with the shortcut for quick voice to quick chat has been resolved.
- Blocked Keys Update: The list of blocked keys has been updated for improved functionality.
Special thanks to Serkan Haşlak, Orhan Kurulan, Takeru, Mert Can Sulupınar, Kamil Andrusz, Julien, Ravi Gouduk, MrSchadenfreude, Necati Koçlu, Aykut Kardaş, Ahmetcan Kavak, Erkan Yenilmez, AudaciousPhysio, ɟnsn⅄, Usman and Nookk for contributing to our latest update with valuable ideas and by reporting bugs, serving as a true inspiration.
improved
fixed
v2.4.2: Performance Improvements
In this update, we've enhanced "Quick Chat/Voice," boosted AI Creativity, and fixed key bugs for a smoother, more intuitive user experience.
Improvements
- Shortcuts for the "Quick Chat" feature are now globally available.
- Performance enhancements have been implemented.
- Improvements related to resource usage.
- The design of the "Quick Voice" feature has been updated.
- New shortcuts have been added to the "Quick Voice" feature.
- A notification sound now indicates when a response is received after using the "Quick Voice Copy" feature.
- The "AI Creativity" feature has been developed further.
Bug Fixes
- The issue of repetitive responses from the AI has been resolved.
- Unintended switching between AI models has been fixed.
- Removed confusing shortcut examples in the new chat screen to reduce complexity.
- The list of prohibited hotkeys has been updated.
Special thanks to Erkan Yenilmez, Ravi Gouduk, Steve Kelley,William Vest,Kamil Andrusz and Takeru for contributing to our latest update with valuable ideas and by reporting bugs, serving as a true inspiration.
new
improved
fixed
v2.4.1: Claude 3: Opus, Sonnet, Haiku
In our latest update, we've introduced new Anthropic models -Claude 3 Opus, Sonnet, Haiku- made several UI and performance enhancements, and fixed a thread deletion bug.
New Features
- Introduced new Anthropic models: Claude 3 Opus, Sonnet, and Haiku.
Improvements
- Aligned sidebar buttons for better interface consistency.
- Optimized backend to reduce latency.
- Enhanced window experience for smoother user interaction.
Bug Fixes
- Fixed an issue with deleting threads from thread settings.
Special thanks to Ivar for contributing to our latest update with valuable ideas and by reporting bugs, serving as a true inspiration.
In our latest update, we've introduced new AI models Groq Mixtral & Llama2, streamlined Apple Sign-In, enhanced window memory and prompt navigation, with a fresh voice wave design and key chat function fixes.
New Features
- Integrated new AI Model: mixtral-8x7b-32768
- Integrated new AI Model: llama2-70b-4096
- Apple Sign In Authentication
- No permission required at start
- Automatic Thread Title
- Memory function for saved window positions (window positions can now be edited)
Improvements
- Prompt menu keyboard scroll controller added
- Fixed prompt menu title length
- Improved quick chat experience with escape key press
- New design for voice wave
Bug Fixes
- Full screen background bug
- Chat input type bug
- Chat input movement with keyboard bug
Special thanks to Mert Can Sulupınar, Guy Cohen, Yakup Öztop, Furkan Kılıç, Maik Frank, Necati Koçlu, Orhan Kurulan, İren Saltalı, Aykut Kardaş, Usman Mirza, Impala, and Julien for contributing to our latest update with valuable ideas and by reporting bugs, serving as a true inspiration.
Load More
→