Claude’s voice mode just grew up.
Until this week, talking to Claude meant talking to Haiku — fast, but built for quick answers, not hard thinking. On July 23, 2026, Anthropic changed that: voice mode now runs Opus and Sonnet too, reaches into your connected tools, and speaks eleven languages. The result is less “voice assistant,” more “thinking partner you can pace the room with.” Here’s what changed and how to use it.
Voice mode is now in beta for every chat user on mobile, desktop, and web. Pick your model mid-conversation, let Claude act in tools like Gmail, Slack, Google Calendar, and Canva (with permission), and switch between 11 languages. Free plan: Haiku + one tool. Paid: all models + all your tools.
01 Why the model swap matters
Anthropic originally put Haiku behind voice mode for one reason: speed. That kept conversations snappy — and shallow. People were already using voice sessions to work through real business problems out loud, and Haiku would keep up but not keep depth. Now the model picker works in voice: choose Opus or Sonnet for hard problems, and Claude uses the fastest version of whichever model you pick so the conversation still flows. It defaults to the last model you used in text chat, so switching between typing and talking doesn’t reset anything.
One design choice worth knowing: Claude’s voice mode takes turns. It listens, pauses to think, then responds. It feels different from assistants that fire back instantly — that pause is the more capable model actually working.
02 What a “thinking out loud” session looks like
03 From talking to doing
The bigger shift: a voice conversation can now end in an action. Once you’ve decided something out loud, ask Claude to do it in your connected tools:
- Running late? “Push my 2pm on Google Calendar back 30 minutes.”
- Finished talking through the pitch? “Turn that into a one-pager in Canva.”
- Walking to the office? “Summarize today’s email and draft replies to the critical ones.”
Claude asks permission before touching any connected tool. You connect tools in Settings → Connectors on mobile, desktop, or web.
04 Eleven languages, one catch
Voice mode now speaks English, French, German, Hindi, Indonesian, Italian, Japanese, Korean, Brazilian Portuguese, and Spanish in both Latin American and European variants — on every plan, including Free. You can switch mid-conversation.
05 Free vs. paid, and the fine print
- Free plan: Haiku in voice, one connected tool, all languages.
- Paid plans: Opus and Sonnet in voice, plus all your connected tools.
- Limits: voice conversations count toward your regular usage limits — a long Opus voice session spends the same budget as heavy Opus text use.
- Where: beta on mobile, desktop, and web; Anthropic says it works best from your phone. Tap the sound-wave icon in the mobile app to start.
Take one decision you’ve been chewing on, go for a ten-minute walk, and talk it through with Sonnet in voice mode. End with: “Summarize what I actually said I wanted.” That summary — your own reasoning, played back — is the feature’s best trick.