How to use voice dictation with Claude Code and Codex in 2026: 9 ways compared

Disclosure: Enjoy is our product. We checked every way below against its maker’s own pages on November 8, 2026, and we say where another way is the better pick, including the free ones that don’t need Enjoy.

The ways to dictate to a coding agent, compared

A prompt for a coding agent is mostly plain language: what’s wrong, what you want, what to leave alone. That makes it a good fit for speaking, and Claude Code, Codex and Antigravity all take voice now. Start with your agent’s own: Claude Code’s and Antigravity’s give the recognizer your project’s words, and Claude Code’s costs nothing on top of your plan. Reach for a dictation app when you want one shortcut that works in every terminal, editor and chat box. The ways below differ most in where your audio goes, and in whether they work over SSH.

Way	Best for	Price, checked November 8, 2026	Runs on	Where speech becomes text
Claude Code’s /voice	Claude Code in a terminal or VS Code	Included in Claude Pro ($20 a month), Max ($100), Ultra ($200), Team and Enterprise	Mac, Windows, Linux	Anthropic’s servers
The ChatGPT app, for Codex	Dictating to Codex, or talking a task through	Dictation in the app; ChatGPT Voice on Plus ($20 a month) and up, at $0.05 a minute of Codex usage	Mac, Windows; Linux in preview	OpenAI’s GPT-Live, for ChatGPT Voice
Antigravity’s voice mode and dictation	Steering Antigravity out loud	Free, with a weekly quota; more with Google AI Pro or Ultra	Mac, Windows, Linux	Google’s Gemini Live, for voice mode
macOS Dictation	Any app on a Mac, with nothing to install	Free	Mac	Your Mac or Apple’s servers; Keyboard settings say which
Windows voice typing	Any app on Windows	Free	Windows	Microsoft’s Azure Speech services
Handy	Free, open-source, offline dictation everywhere	Free	Mac, Windows, Linux	Your computer
Superwhisper	On-device models, and plugins for coding agents	Free; Pro $8.49 a month, $84.99 a year or $249.99 once	Mac, Windows, Linux, iPhone, Android	Your computer, or a cloud model you choose
Wispr Flow	Dictation that reads names from your editor, on desktop and phone	Free with a weekly word limit; Pro $15 a month, or $12 billed yearly	Mac, Windows, iPhone, Android	Wispr’s cloud
Enjoy	Every lab’s agent, with your voice kept on your computer	A monthly subscription, free for 7 days (pricing)	Mac, Windows, Linux; not the web or iPhone app	Your computer

Prices are per month as each maker lists them, before tax. Claude Code needs a paid Claude plan, and Claude’s pricing has Pro at $20 a month, or $17 billed yearly. Codex comes with every ChatGPT plan, and its pricing lists Free at $0, Go at $8, Plus at $20 and Pro from $100. A dictation app types into whatever has focus, so it works with any agent, including the ones in this table.

Claude Code’s /voice

Best for: Claude Code users on a Claude plan, in the terminal or VS Code.

Claude Code has dictation built in. Run /voice, then, in Anthropic’s words, “either hold a key while you speak or tap once to start and again to send.” Hold mode is push-to-talk on Space, and your words appear in the prompt as you speak, so you can mix talking and typing. /voice tap starts with one tap and sends with the next, submitting “automatically when the transcript is at least three words long.” You can rebind the key, set autoSubmit, and dictate in any of 20 languages through the language setting.

It’s built for code: “Transcription is tuned for coding vocabulary,” and “Common development terms like regex, OAuth, JSON, and localhost are recognized correctly, and your current project name and git branch name are added as recognition hints automatically.” It works in the VS Code extension too. And it doesn’t eat into your plan: “Transcription does not consume Claude messages or tokens and does not count toward the limits shown in /usage.”

Price: included in Pro, Max, Ultra, Team and Enterprise. Pro is $20 a month, or $17 billed yearly, Max $100 and Ultra $200. Claude’s Free plan doesn’t include Claude Code.

Watch for: “Voice dictation streams your recorded audio to Anthropic’s servers for transcription. Audio is not processed locally.” It needs a Claude.ai sign-in and “is not available when Claude Code is configured to use an Anthropic API key directly, Amazon Bedrock, Google Cloud’s Agent Platform, or Microsoft Foundry.” It “does not work in cloud sessions or SSH sessions,” and in WSL it needs WSLg. In tap mode, “Recording also stops automatically after 15 seconds of silence or two minutes total.”

The ChatGPT app, for Codex

Best for: Codex users who want to dictate a prompt, or talk a task through out loud.

The ChatGPT desktop app, where Codex runs on your computer, has two kinds of voice. For dictation, its prompting guide says to “select Dictate in the composer, then start talking,” and “ChatGPT transcribes your speech into the composer so you can review and edit it before sending the prompt.” For a conversation, ChatGPT Voice, “Powered by GPT-Live,” lets you “talk through ideas and coordinate tasks in Chat, Work, and Codex.” It “can start separate tasks for longer work, check existing tasks, and send follow-up instructions,” and on macOS, with Screen context on, it can take an appshot of your frontmost window when you say “Take a look at this.”

In the terminal, the Codex CLI has /voice, or F8, “When your CLI build and session support voice.” Its configuration reference calls it “the experimental /voice command.”

Price: ChatGPT Voice “is available in the ChatGPT desktop app with ChatGPT Plus, Pro, Business, Edu, and Enterprise plans.” Plus is $20 a month and Pro from $100. Codex’s pricing says “Voice in Desktop uses your existing Codex usage budget at $0.05 per minute,” and “Voice and tasks share your plan’s usage limits.”

Watch for: “Only one voice chat can be active across the ChatGPT desktop app at a time,” “Availability also depends on rollout status and workspace settings,” and “ChatGPT Voice in Desktop isn’t available via API key.” The app runs only Codex, on macOS and Windows, with Linux “available in preview.”

Antigravity’s voice mode and dictation

Best for: Antigravity users who’d rather talk than type.

Google’s agent has both kinds too. Its dictation types into the message box: click the microphone or press Ctrl+M in the app, or run /voice dictate or press F5 in the CLI, and “Antigravity gives the speech recognizer your recent conversation and the text around your cursor, so project-specific terms are more likely to come through correctly.” Its voice mode, “enabled by Gemini Live,” is a conversation: “Talk with the agent out loud, and hear it answer in real time.” You can give tasks, steer work as it runs, and answer permission prompts out loud. In the app, clear the message box and click Start Voice mode; in the CLI, from version 1.5.0, run /voice live or press F6. Voice mode arrived with Antigravity 2.29.0, which began rolling out on November 3.

Price: Antigravity’s plans give accounts without a paid Google plan “A meaningful quota, refreshed weekly,” and Google AI Pro and Ultra raise it. The voice mode page names no plan.

Watch for: “Voice mode might also not be available for your account yet,” and its button is hidden once you’ve reached a usage limit. “In the Antigravity App, voice mode is available only in local conversations,” and “You can’t approve destructive actions by voice,” such as rm, sudo or a force-push. Over SSH, the CLI needs agy mic-serve running on your own computer.

macOS Dictation

Best for: any app on a Mac, with nothing to install.

“With Dictation, you can speak to enter text anywhere you can type it,” Apple says, and that includes the terminal Claude Code or Codex runs in. Turn it on under System Settings → Keyboard → Dictation, then “Press the Microphone key if available in the row of function keys on your Mac, use the Dictation keyboard shortcut, or choose Edit > Start Dictation.” “You can dictate text of any length without a timeout,” and “On a Mac with Apple silicon, you can still use the keyboard while you speak.” To see where your words are processed, Apple says to check Keyboard settings for whether dictation is “processed on your device and not sent to Siri servers.”

Price: free.

Watch for: “Dictation stops automatically when no speech is detected for 30 seconds,” and “Dictation is not available in all languages or regions, and features may vary.” “Ambiguous text is underlined in blue,” as when you get ‘flour’ for ‘flower.’ Check file and function names before you send.

Windows voice typing

Best for: any app on Windows, with nothing to install.

“With voice typing, you can enter text on your PC by speaking,” Microsoft says. Put your cursor in Windows Terminal, PowerShell or your editor, press Windows key + H, and wait for the Listening… alert before you speak. It can add punctuation for you, and to stop, you say a command like “Stop listening” or press the microphone button.

Price: free.

Watch for: “To use voice typing, you’ll need to be connected to the internet, have a working microphone, and have your cursor in a text box,” because “Voice typing uses online speech recognition, which is powered by Azure Speech services.”

Handy

Best for: free, offline dictation in every app, on any computer.

Handy is, in its README’s words, “A free, open source, and extensible speech-to-text application that works completely offline”: “Press a shortcut, speak, and have your words appear in any text field. This happens on your own computer without sending any information to the cloud.” You hold the shortcut to record or tap it to toggle, and choose Whisper models or Parakeet V3, a “CPU-optimized model with excellent performance and automatic language detection.” It’s under the MIT License, on macOS, Windows and Linux, and its latest release, 0.10.0, came out on November 2.

Price: free. “Accessibility tooling belongs in everyone’s hands, not behind a paywall.”

Watch for: it “has some known issues,” its README says, including “Limited support for Wayland display server,” where it needs wtype or dotool to type. “Shortcuts that include the fn (Globe) key only work on Apple keyboards,” so pick another if you use a third-party keyboard.

Superwhisper

Best for: on-device models you choose, and a Mac plugin that lets you answer your agent by voice.

Superwhisper offers “Voice to text that works in any app,” and “Superwhisper works offline, so you can transcribe anytime.” It runs on Mac, Windows, iPhone and Android, with a Linux app that says “Process locally,” and its homepage pitches it for “Cursor, Claude Code, Open Code, Amp, Codex, or any other agentic coding app.” On a Mac, its coding agents integration goes further: for Claude Code, OpenCode, Codex, Cursor, Pi, Grok and DeepSeek, “Superwhisper pops up when your agent needs your input or finishes working, and you answer by voice instead of typing,” and you can approve or deny tool calls by speaking.

Price: the free plan includes “Unlimited use of Whisper models.” Pro is $8.49 a month, $84.99 a year or $249.99 for life, and adds “Use your own AI API Keys” and “Unlimited use of Cloud and Local AI models.” New users get Pro features “for 3,000 words for free,” and there’s a “30 day refund available for all plans.”

Watch for: “Coding agent integrations are currently available on macOS only,” and “Intel Macs work best with Cloud models. Offline models only run really well on Apple Silicon macs.”

Wispr Flow

Best for: dictation on desktop and phone that picks up names from your editor.

Wispr Flow offers “Speech-to-text in any app, 100+languages” and “Dictation on Mac, Windows, iOS, and Android.” For code, its IDE guide says “Variable recognition picks up names visible in your editor; file tagging attaches files you mention in a supported AI chat.” Variable recognition works in Cursor, Windsurf and VS Code, and file tagging in Cursor’s and Windsurf’s chat panels. You hold Fn on a Mac or Ctrl+Win on Windows and speak, and “Flow pastes long dictations into Claude Code and Codex CLIs in plain-text parts.”

Price: free, with “Limited words per week”: on Mac and Windows, its docs put the soft cap at 2,000 words and the hard cap at 5,000. Pro is $15 a user a month, or $12 billed yearly, with “Unlimited dictations,” and Growth is $23, or $18 yearly.

Watch for: “Wispr Flow processes dictated audio in the cloud to turn speech into text,” and its privacy overview says Improve the model for everyone “is on by default for Free and Pro,” though you can turn it off. “There is no native Linux application,” and for “WSL windows, Linux VMs, remote SSH sessions,” it says, “Flow transcribes, but you must use Paste last transcript.”

Enjoy

Best for: speaking to every lab’s agent on Mac, Windows or Linux, with your voice kept on your computer.

Enjoy runs Claude Code, Codex, Grok Build, Antigravity, Muse Code, and local models through OpenCode, on Mac, Windows and Linux, on the plans you already have, and in the desktop app one microphone button in the message box works the same for all of them. The first time, you choose how it hears you. On a Mac, that’s Apple Speech, which comes with macOS, or Qwen3-ASR 0.6B (a 1.0 GB download) or Qwen3-ASR 1.7B (2.5 GB, the most accurate), speech models built on a language model. On Windows and Linux, “Enjoy offers Qwen3-ASR 0.6B only,” which runs on your computer’s processor. Your words appear as you speak, and pressing Enter sends the message as soon as the last words are in. You can also hold a key and speak in any other box in Enjoy, such as a search or a note: fn on a Mac, Right Ctrl on Windows and Linux. Either way, “speech turns into text on your computer. Your voice isn’t sent to Enjoy, to your agent, or anywhere else. Only the message you send is.”

Price: Enjoy is one monthly subscription, and the first seven days are free, with a card to start. It includes your phone and any browser and one computer; each other computer is $9.99 a month, and each teammate you invite needs a subscription of their own. See pricing.

Watch for: “Dictation is in the computer app.” On a Mac, Apple’s speech recognition “needs macOS 26 or later,” and the downloadable models “need Apple silicon (M1 or later).” On Windows and Linux, the model “needs a 64-bit Intel or AMD processor from about 2013 or later, with at least four threads.” The web app doesn’t dictate, and “In the iPhone app, use the microphone on the keyboard instead.” It types only into Enjoy, not your terminal or editor. Enjoy isn’t open source, and it’s a monthly subscription on top of your AI plans once the free week ends.

When to stay with what’s built in

If you use one agent, start with its own voice. Claude Code’s /voice knows coding words, your project’s name and your branch, and costs nothing on top of your Claude plan. Codex users already have Dictate in the ChatGPT app, and Antigravity users have dictation and voice mode in the app and the CLI. If you only want words in a box now and then, the dictation built into macOS and Windows is already on your computer.

Look further when you want one shortcut for every agent and every app, or your audio to stay on your computer. Handy is free and offline everywhere, Superwhisper adds on-device models and agent plugins on a Mac, and Wispr Flow picks up names from your editor, if cloud processing is fine for your work. Enjoy fits when you already run several labs’ agents in it.

How we compared

We read Anthropic’s Claude Code docs on voice dictation and VS Code, Claude’s pricing, OpenAI’s Codex docs on ChatGPT Voice, prompting, slash commands, configuration, the desktop app and pricing, Antigravity’s docs on voice mode, features and plans, Apple’s and Microsoft’s dictation guides, Handy’s homepage and GitHub page, Superwhisper’s homepage, Linux page and docs, Wispr Flow’s pricing and docs, and Enjoy’s own help on November 8, 2026, and every one is linked below. We quote each maker’s own words for its limits and leave out what it hasn’t documented. We didn’t test accuracy, so we don’t rank it. These apps change weekly, so we recheck this guide against its sources at least once a month, and the date at the foot of the page says when we last did.