Claude Voice Mode Gets Smarter: Opus, Sonnet, Your Email and Calendar, and More Languages (2026)

What Anthropic actually changed

On July 23, 2026, Anthropic announced a meaningful upgrade to Claude's voice mode, the hands-free feature that lets you talk to Claude out loud and hear it answer. In Anthropic's words: "Voice mode now runs on Claude's more capable models and reaches the tools you've connected mid-conversation."

Three things changed at once. First, voice mode now runs on Claude's stronger models, Opus and Sonnet, not just the lightweight Haiku it used before. Second, it can reach the connectors you have linked to Claude, like your email and calendar, in the middle of a spoken conversation. Third, it speaks in many more languages. Together, that turns voice mode from a quick way to ask Claude a question into something you can actually think and work through out loud.

None of this is a separate app or a hidden beta. It is the same voice mode, built into the Claude apps you already use, now noticeably more capable. Here is how Anthropic announced it:

Straight from Anthropic:

From Haiku-only to Opus and Sonnet

Until now, Claude's voice mode ran only on Haiku, Anthropic's fastest and lightest model. Haiku is wonderful for snappy back-and-forth, but it was never meant to grind through a thorny analysis or a multi-step plan. So the old voice mode was great for quick questions and awkward for anything that needed real thinking.

This update changes that. Voice conversations can now use the same models you rely on in text chat, including Claude Sonnet and Claude Opus. You can switch models mid-conversation from the model selector: stay on Haiku for a fast chat, then jump to Sonnet or Opus the moment a question gets hard. (One caveat: Claude's Fable model is not available in voice mode yet.)

Why does this matter? Because it removes the reason you used to drop out of voice and go back to typing. When the model behind the voice is as capable as the one in your text window, you can talk through the hard problems, a strategy, a tricky email, a piece of code, instead of just asking for the time or the weather.

Now it can reach your email and calendar, while you talk

The second big change is about connection. Claude can be linked to the tools you use: your email, your calendar, your documents, and other connectors. Before this update, voice mode could not touch any of that; it was a conversation sealed off from your actual day. Now it can reach those connected tools in the middle of a spoken conversation.

In practice, that sounds like this: "What's on my calendar tomorrow afternoon?" "Summarize the last email from Priya." "Move my 3pm to Thursday." Claude pulls from what you have connected, reads the answer back, and, with your go-ahead, can act on it. Anthropic's own examples include turning a rambling spoken conversation into a tidy one-page pitch, or shuffling your calendar around, all by voice.

This is the difference between a voice that only talks and a voice that can also fetch and do. Here is how Anthropic put it:

Think out loud, in many more languages

Anthropic's framing for the update is worth repeating: "Talk through the hard problems out loud, in many more languages." The "out loud" part is the point. With a capable model behind it, voice stops being a shortcut for simple commands and becomes a way to reason, to talk yourself through a decision while Claude keeps up.

The language expansion widens who can do that. At this update, voice conversations are supported in 11 languages: English, French, German, Hindi, Indonesian, Italian, Japanese, Korean, Brazilian Portuguese, and Spanish for both Latin America and Spain. And you can switch languages mid-conversation, which is genuinely useful if you live or work between two of them.

For non-English speakers especially, this is the update that makes voice mode feel native rather than bolted on. You can think in Japanese, switch to English to draft a message, and switch back, without ever opening a settings menu.

How to turn on and use Claude voice mode

It is built into the apps you already have. On desktop or the web, open a new chat and click the sound-wave icon in the lower-right corner, then start speaking; your words appear as you talk, and Claude replies when you pause. On mobile (iOS and Android), tap the voice icon in the message box, pick a voice, and go. Tap "Stop" to leave voice mode.

To use the new models, tap the model selector while you are in voice mode and switch to Sonnet or Opus (or back to Haiku for speed). To set the language Claude speaks, go to Settings > General > Voice > Language; note that changing your app's display language does not change your voice language, so set it here. In a noisy room, turn on push-to-talk so Claude only listens while you hold the button.

Voice mode is available on every plan, Free, Pro, Max, Team, and Enterprise, and usage counts toward your normal limits. To reach your email and calendar by voice, connect those tools to Claude first; once they are linked, voice mode can use them in-conversation.

Talking to Claude vs. doing your work by voice

Here is the honest limit, the same one every voice assistant runs into. Claude voice mode is a superb conversation partner, now smarter, more connected, and multilingual. But it is a conversation that lives inside the Claude app. Ask it to check your calendar and it will; ask it to fire off three Slack replies, tweak the paragraph you just wrote in Notion, and rename a file in Finder, and you have left what a chat assistant is built to do.

Most work does not happen inside one app. It happens across Gmail, Slack, Notion, your browser, your calendar, and your code editor, and it is mostly doing, not discussing. A voice that is brilliant inside one window is a brilliant chatbot. A voice that can type into any field, rewrite the text already on your screen, and take actions across all your apps is something closer to an operating system.

That is the line between talking to a computer and getting your computer to do the work, a line we have mapped out before in our piece on voice agents vs. dictation. Claude's update makes the conversation smarter. The next question is what that conversation can actually do across everything else you use.

Where VoiceOS fits

VoiceOS is that system-wide voice layer for your Mac and Windows computer. It is not sealed inside one app. You press your trigger, speak naturally, and VoiceOS turns your words into the right output wherever your cursor is: a polished Slack message, a clean email, or a prompt for Claude itself, in the tone of the app you are in.

And it goes past talking into doing. With Agent Mode you can ask about what is on your screen, reshape text you have already written, and take real actions across connected apps like Gmail, Slack, and Google Calendar, with a confirmation step before anything important goes out. Claude is even one of the 20+ tools VoiceOS plugs into, so the two work together: talk with Claude when you want to think, and talk through VoiceOS when you want it done.

Anthropic's upgrade is one more sign the whole industry is converging on the same idea we started with: you should be able to talk to your computer and have it understand and act. If that future excites you, the fastest way to feel it across your real work today is to start speaking to your own machine.

Frequently Asked Questions (FAQ)