2 min read AI-generated

Claude's Voice Mode Grows Up: Opus, Sonnet and Haiku on Command

Copy article as Markdown

Anthropic beefed up Claude's voice mode. You can now pick between Opus, Sonnet and Haiku — and while it talks, Claude can reach into Gmail, Calendar, Slack, Canva and Notion. The point: it doesn't just chat, it gets things done.

Featured image for "Claude's Voice Mode Grows Up: Opus, Sonnet and Haiku on Command"

A few weeks after OpenAI reworked its voice mode, Anthropic is making its move. On Thursday the company upgraded Claude’s voice mode — and the jump is bigger than it looks at first glance.

Finally, a model picker

Until now, the voice mode Anthropic launched last year ran purely on Haiku. Fast to respond, sure — but simply too thin for anything complex. That changes now: you can choose between Opus, Sonnet and Haiku. By default, voice mode picks up whatever model you last used in text chat and runs its fastest version.

It sounds like a small toggle, but it makes a real difference. With a stronger model behind it, voice mode suddenly holds up in longer conversations: feedback on your communication style, rehearsing a client pitch, or brainstorming product market research.

It doesn’t just talk — it acts

Here’s the genuinely interesting part: Claude’s voice mode can tap into other apps — Gmail, Google Calendar, Slack, Canva and Notion. So you can move a meeting slot, draft an email or create a Notion doc, all by voice. And this is exactly where it pulls away from OpenAI: their voice mode got a conversational polish, but still can’t use tools to actually get work done.

There’s multilingual support too, in beta. It currently covers English, French, German, Hindi, Indonesian, Italian, Japanese, Korean, Brazilian Portuguese and Spanish — though you still have to set the language manually.

It’s available to all users, across platforms, as a beta. Free users stay capped at Haiku and a single connected app.

My take

One honest caveat belongs here: Anthropic didn’t touch the underlying voice model and won’t say anything about its voice stack. So if you were hoping for smoother interruptions or a more natural back-and-forth — like OpenAI’s last update — you won’t find them. The progress is elsewhere: in the intelligence behind it and the ability to kick off real work by voice. And that fits Anthropic’s line perfectly — Claude isn’t meant to be a chat buddy, but an assistant that does things. Here, speech is just the input method.


Sources: