Grok Build v1.0.0 Changelog

This is the first major release of Grok, and it’s all about making the tool feel faster and less frustrating to use. It matters to anyone who uses Grok to write code or manage projects, especially if you work with large files or use remote servers.

What's new

• Dashboard summaries: The main screen now shows a quick recap of what the AI did in its last turn, so you don't have to scroll up to remember.
• Better extensions menu: Add-ons are now sorted alphabetically, and you can collapse (hide) sections to keep things tidy.
• Smarter startup: If you open Grok outside of a specific coding project folder, it won't bother asking you which project you want to work on.
• Dedicated feedback box: Typing /feedback now opens a special window to send your thoughts, rather than just typing into the chat.
• Automatic dark/light themes: The app can now detect your computer's color theme automatically, even when you're using remote connections like SSH or terminal multiplexers like tmux (tools for managing multiple terminal windows).
• Wrapping tables: If you're in a narrow window, text inside tables will wrap to the next line instead of getting cut off.
• Full script previews: When Grok asks for permission to run a script, it shows you the whole thing. If it's a really long script, you can press Ctrl-F to expand it and read it all.

Improvements

• Saves computer memory: Duplicating very large chat sessions no longer eats up a massive amount of your computer's memory (RAM).
• Faster exits: Closing out of an empty session is now instant, even if your internet connection is slow.

Fixes

• Fixed image corruption: Tools that send pictures back to Grok no longer drop or ruin large screenshots.
• Fixed startup crashes: Starting Grok in a "sandbox" (a safe, isolated space) on folders with lots of blocked files no longer breaks.
• No more lost messages: Smashing the "send" button really fast won't accidentally delete the messages you lined up to send.
• Better task canceling: Hitting Esc or "stop" now actually stops background tasks from restarting the AI after you cancel them.
• Fixed login issues: The app won't accidentally skip logging in if you have a bad API key (a passcode for the software) saved in your settings.
• Unfrozen menus: You can now use the model picker and command search bar while reviewing an AI's plan, without the screen freezing.
• Consistent keyboard controls: The Tab and Esc keys now work the same way across all pop-up cards (like questions or cancel buttons).
• Smoother navigation: Typing /new from the main screen now correctly takes you back to the main screen if your chat is empty.
• Fixed code restores: Restoring your code from a backup no longer freezes up if you're using a really large or very shallow git repository (a save-history system for code).
• Smarter remote resumes: When resuming a remote session, it only brings back your chat history unless you specifically tell it to also restore your code files.
• Fixed text copying: When highlighting and copying languages like Chinese or Japanese with your mouse, it no longer misses characters at the very beginning or end of what you selected.
• Cleaner error messages: When the server messes up, you now get a clean, readable banner instead of a messy dump of computer code (JSON).
• Easier exiting: Typing "exit" or "quit" on the main screen now actually closes the app.
• Accurate mode indicators: The icon showing what mode you're in (like planning or asking) no longer gets confused after you resume a session.
• Better menu actions: Pressing Enter in the slash-command menu now actually runs the command you highlighted, and deleting a session correctly takes you back to the main screen.
• Tougher during outages: Grok now tries harder to fix server errors behind the scenes instead of just failing.
• Cleaner waiting screen: Messages waiting in line now stay visible while the AI is thinking, and you can reorder things like images and commands in the queue.
• No more interrupting recaps: The AI no longer tries to summarize things while it's still working on a task.
• Fixed text wrapping: Error messages from the /btw command now wrap fully so you can read the whole thing.

Breaking changes

No breaking changes in this release.