Claude’s Voice Mode Arrives on Opus and Sonnet — And Learns to Use Your Apps

Anthropic has expanded Claude’s voice mode to its two most capable models, Opus and Sonnet, moving the feature well beyond the quick back-and-forth it launched with. At the same time, the company is letting voice mode reach into the apps people already use every day — including Gmail, Slack, and Canva — turning a talking assistant into something closer to a hands-free operator for real work.

Key takeaways

• Voice mode was previously limited to Haiku, Anthropic’s fastest but least powerful model — great for speed, not for depth.
• It now runs on Opus and Sonnet, so spoken conversations can tackle genuine business problems, not just quick lookups.
• Claude can now act inside connected apps like Gmail, Slack, and Canva while you talk to it.
• The shift reflects how people actually used voice — for deep, multi-step work rather than casual queries.

From quick answers to real conversations

When Anthropic first launched voice mode, the goal was simple: answer short questions with as little delay as possible. Haiku was the natural fit — it is fast and responsive, keeping the conversation snappy. But according to the company, users almost immediately pushed the feature past its original design, using it to talk through real business problems. Haiku kept things quick, but, as Anthropic put it, not always deep.

Opus and Sonnet change that equation. These are Anthropic’s more powerful reasoning models, built to hold longer, more nuanced threads of thought. Bringing them into voice means a spoken exchange can now carry the same weight as a typed one — working through a strategy, drafting a plan, or reasoning across several steps without losing the thread.

Voice that reaches into your tools

The more consequential change may be the app connections. By extending voice mode into services like Gmail, Slack, and Canva, Anthropic is letting Claude do more than describe an answer — it can help act on it. That points toward a future where you speak a request and the assistant drafts the email, checks the message thread, or pulls up the design, all without you touching a keyboard.

It is a meaningful step in the broader industry race to make AI assistants genuinely useful in the flow of work, rather than a separate window you visit for one-off questions.

Why it matters

Voice is quietly becoming the interface where the assistant wars will be won. A model that can both reason deeply and take action inside your apps — entirely by voice — is far closer to a capable digital coworker than a novelty. For businesses, the takeaway is clear: the gap between “asking an AI” and “having an AI do it” keeps shrinking.