Claude’s voice mode is now available for Opus and Sonnet
The Verge Terrence O’Brien ● Covered by 8 sources
Claude's voice mode now runs on Opus and Sonnet, not just the lightweight Haiku model. Translation: talking to Claude just got a lot smarter.
Based on reporting by The Verge, Terrence O’Brien — 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 admitted something interesting this week: people weren't using Claude's voice mode the way the company expected. When voice mode launched earlier this year, it ran exclusively on Haiku, Anthropic's fastest and cheapest model, built for snappy back-and-forth rather than heavy lifting. The pitch was simple — ask a quick question, get a quick answer, move on with your day.
But users had other plans. Instead of treating voice mode like a glorified Siri replacement, people started dictating their way through actual business problems, according to Anthropic's own blog post. Haiku, by design, kept things fast and shallow. It wasn't built to reason through anything complicated, which meant a growing chunk of Claude's voice users were bumping into the limits of a model that was never meant to carry that weight.
So now Opus and Sonnet — Anthropic's more capable, more expensive models — are available in voice mode too. That's a meaningful shift. Opus in particular is the model Anthropic reserves for its hardest reasoning tasks, and putting it behind a microphone icon changes what voice mode is actually for. This isn't about faster answers anymore; it's about talking through messy, multi-step problems out loud and having the model keep up.
Anthropic is also widening where voice mode can reach, plugging it into Gmail, Slack, and Canva. That's a telling combination — productivity tools, not entertainment apps. It signals Anthropic sees voice as less a novelty feature and more an interface for getting real work done, which lines up neatly with the behavior it noticed from users in the first place.
The bigger story here might be how fast Anthropic reacted. Voice mode is barely a few months old, and the company already rebuilt its foundation because users pushed it somewhere it wasn't designed to go. That kind of quick pivot is becoming a pattern across AI products generally — ship something narrow, watch what people actually do with it, then rebuild around the real use case instead of the assumed one.
My take — AI-written commentary, not fact-checked reporting
I'll say the quiet part: shipping voice mode on Haiku first always felt like a cost-saving move dressed up as a product decision, and the fact that users immediately outgrew it proves the point. Good on Anthropic for listening and upgrading fast instead of defending the original choice, but the real lesson is that people don't want a toy voice assistant — they want their actual model, just spoken aloud.
Read more about this at: The Verge