What this covers
If you use ChatGPT every day and your hand keeps drifting to the trackpad, you’re paying a tax in seconds-per-interaction that compounds. The shortcut menu is real and most of it is forgettable; this is the short list of shortcuts that pay off, with a few opinionated notes on which ones to memorize first. Plus a couple of OS-level tricks that aren’t strictly ChatGPT but slot in next to the workflow.
Who this is for
Power users who spend 30+ minutes a day in ChatGPT on a laptop. If you’re a casual user, three shortcuts will do (new chat, copy last response, stop generating); skip the rest. If you’re in there all day, the full list is worth the 20 minutes to learn.
When to reach for it
- You start dozens of chats a day and the ”+ New chat” click is friction.
- You frequently re-copy the model’s last response into another tool.
- You generate long answers and want to interrupt early when you can see it’s going wrong.
- You’re navigating between many sidebar conversations and want to search by keyboard.
Before you start
- These shortcuts work in the web app and desktop app. Mobile is touch-only; skip this guide for phone use.
- On Mac, the modifier is Cmd. On Windows / Linux, it’s Ctrl. The patterns are otherwise identical.
- Open the in-app shortcut menu once to see your current bindings — OpenAI occasionally changes them, and a third-party browser extension might be intercepting some.
- Some shortcuts (e.g. Cmd/Ctrl+Shift+;) conflict with browser or system bindings on certain setups. If one doesn’t work, check your OS / browser keymap.
The shortcuts that pay off
-
New chat —
Cmd/Ctrl + Shift + O. This is the highest-payoff one. Skipping the ”+” click adds up to dozens of saved interactions per week. Memorize first. -
Copy last response —
Cmd/Ctrl + Shift + C. Especially useful when you’re pasting outputs into a doc, terminal, or another app. Skips the “copy” button at the bottom of each message. -
Open shortcut menu —
Cmd/Ctrl + /. Your quick-reference. If you forget any of the others, this is the universal fallback. -
Focus chat search —
Cmd/Ctrl + Shift + ;. Pulls up the sidebar search field. Combined with descriptive chat titles, this replaces scroll-hunting through hundreds of conversations. -
Stop generating —
Esc. Worth its weight when you can see by the third sentence that the model misread your prompt. Saves model time, saves your time. -
Newline without sending —
Shift + Enter. Not strictly a shortcut, but worth listing: Enter sends, Shift+Enter inserts a newline. Critical for writing multi-paragraph prompts. -
Set custom instructions / settings — typically accessible from the shortcut menu (see #3). The exact binding has shifted; check your menu.
Cheat-sheet recap
Cmd/Ctrl + Shift + O New chat
Cmd/Ctrl + Shift + C Copy last response
Cmd/Ctrl + / Open shortcut menu
Cmd/Ctrl + Shift + ; Focus chat search
Esc Stop generating
Shift + Enter Newline without sending
If you only memorize three: new chat, copy last response, stop generating.
Quality check
- Hold yourself accountable: for a week, every time you reach for the trackpad, ask if there’s a shortcut. If yes, undo and use it. The habit hardens fast.
- If a shortcut doesn’t fire, open Cmd/Ctrl+/ and verify the binding — yours might differ from this list if you have an older app version.
- Watch for browser extensions intercepting keys (Vimium, Surfingkeys, password managers). Disable on chat.openai.com if needed.
- On Linux, some window managers grab Cmd/Ctrl+Shift sequences globally — check your DE’s keybinding settings.
How to reuse this workflow
- Pair shortcuts with descriptive chat titles. The keyboard-search shortcut is useless if your sidebar reads “Untitled chat (1)” through “Untitled chat (47).”
- For repeating tasks, combine
Cmd/Ctrl + Shift + O(new chat) with a saved prompt template — paste, edit a slot, fire. - Teach a colleague these six shortcuts and watch their usage speed double. The compounding is real.
Stacking sequence
Cmd/Ctrl + Shift + O (new chat) → paste prompt template → edit slot → Enter → watch output → Esc if it’s going wrong → otherwise wait → Cmd/Ctrl + Shift + C (copy) → paste into target tool.
Common mistakes
- Trying to memorize all of them at once. Pick the top three, use them for a week, then add the rest.
- Forgetting Shift+Enter for newlines and ending up with one-line prompts that ramble. Multi-paragraph prompts are clearer.
- Not using Esc when the model is clearly going wrong — you sit there watching a wall of bad text, then delete it. Esc and re-prompt.
- Relying on Cmd/Ctrl+Shift+; for search and never titling chats. Search needs something to match against.
- Assuming the bindings are identical to a year ago — OpenAI has changed them. Recheck the in-app menu when you upgrade.
- Conflating “the chat input has focus” with “the page has focus.” If your shortcut isn’t firing, click once outside the textarea and try again.
FAQ
- Are there shortcuts for switching models?: There’s a model picker dropdown reachable by keyboard if you tab through the chat header, but no first-class hotkey at time of writing. The fastest path is to set a default model and switch only when needed.
- Can I remap shortcuts?: Not within ChatGPT itself. You can use OS-level remapping tools (Karabiner on Mac, AutoHotkey on Windows) for personal bindings, but verify they don’t intercept the in-app ones.
- Do shortcuts work in the desktop app the same way?: Yes, with minor differences — the desktop app sometimes adds its own global hotkey for “summon ChatGPT” that the web doesn’t have.
- What about voice input as a “shortcut”?: On desktop, system-wide dictation (Mac: Fn-Fn, or your configured key) into the chat textarea is faster than typing for many people. Worth pairing with the keyboard workflow.