Claude’s Voice Mode Gets a Brain Boost and Connects to Your Apps

Anthropic is giving its voice mode a serious upgrade. Starting today, Claude’s speech feature can tap into the company’s smarter Sonnet and Opus models, not just the basic Haiku system. That means complex questions—like asking it to summarize a long email thread or walk through a multi-step project—won’t trip it up as easily.

If you pay for Claude, the voice tool will automatically use whichever model you last selected for text chats. You can also swap between Haiku, Sonnet, or Opus mid-conversation using the model picker. Anthropic says the system picks the fastest version of your chosen model to keep things snappy. And here’s the big addition: voice mode can now pull context from apps like Gmail and Slack, as long as you give permission. So you could ask Claude to read your latest messages aloud or help draft a reply.

There are still trade-offs. Claude’s voice mode works on a turn-based system—it listens, pauses to think, then responds. That makes it feel less like a real-time conversation than OpenAI’s GPT-Live, which can talk and listen at the same time. It also can’t automatically detect language switches mid-chat; you’ll need to tell it you’re switching or pick the new language in settings. On the plus side, Anthropic added support for Indonesian and several other languages. The update rolls out in beta today across desktop, mobile, and web. Free users get one voice connection and are limited to Haiku, but paid subscribers can unlock the full experience.