Anthropic updates Claude voice mode with more capable models
TechCrunch Ivan Mehta ● Covered by 8 sources
Claude's voice mode now lets you pick Opus, Sonnet, or Haiku instead of being stuck with one model. It can also hook into Gmail, Slack, Notion and Calendar to actually do stuff, not just chat.
Based on reporting by TechCrunch, Ivan Mehta — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Anthropic quietly leveled up Claude's voice mode this week, and the timing is no accident. OpenAI just refreshed ChatGPT's voice experience with a new conversational model family, and Anthropic clearly wasn't going to let that sit unanswered. The update, announced Thursday, lets users choose between Opus, Sonnet, and Haiku when talking to Claude out loud, rather than being locked into the single Haiku-powered setup that launched last year.
That original voice mode was snappy but shallow. Fine for quick questions, useless for anything that required real reasoning. Now Claude defaults to whatever model you were last using in text chat, running its fastest variant, which means the assistant can carry a longer, more substantive conversation. Anthropic points to use cases like rehearsing a client pitch out loud, getting live feedback on how you communicate, or talking through product research instead of typing it all out.
The more interesting piece, though, is what Claude can actually touch. Voice mode now connects to Gmail, Google Calendar, Slack, Canva, and Notion, so you can ask it to move a meeting, draft a message, or spin up a document without switching apps. That's a meaningful gap between Anthropic and OpenAI right now — ChatGPT's voice update focused on making conversations feel more natural, but it still can't reach into your tools and do the work. Claude can, at least in this beta.
Anthropic also expanded language support earlier this year, adding beta coverage for French, German, Hindi, Indonesian, Italian, Japanese, Korean, and both Brazilian Portuguese and Spanish, alongside English. You still have to tell it which language to use, so it's not automatically detecting anything yet.
One thing didn't change: the voice itself. Anthropic hasn't touched the underlying voice model or said what stack powers it, so anyone hoping for smoother interruption handling or a more natural back-and-forth cadence, the kind OpenAI has been chasing, won't find it here. Free users, meanwhile, get Haiku only and a single connected app, so the full version of this update is really aimed at paying customers.
My take — AI-written commentary, not fact-checked reporting
I think Anthropic made the smarter bet here by skipping the voice-quality arms race and going straight for utility — an assistant that can actually touch your calendar and inbox beats one that just sounds more human while doing nothing. OpenAI's version might feel more pleasant to talk to, but pleasant small talk isn't why people pay for these things.
Read more about this at: TechCrunch
