Claude Voice Mode Now Runs on Opus and Sonnet - and Can Take Actions in Gmail, Slack, and Notion
While OpenAI updated ChatGPT's voice with more natural conversation, Anthropic went a different direction: making Claude voice mode smart enough to actually get work done across your apps.
Anthropic upgraded Claude voice mode today, adding support for Opus and Sonnet models for the first time. Claude voice mode previously ran only on Haiku - fast, but limited to simpler tasks. Now paid users can choose the same model tier they use in text chat, and voice mode picks the fastest version of their last-used model by default. Voice mode can also switch models mid-conversation.
Claude Voice Can Now Control Your Apps
Claude voice mode now works with app connectors. Users can speak to Claude and have it update a calendar event in Google Calendar, draft an email in Gmail, create a Notion document, or post to Slack. Free users get access to one connected app with the Haiku model. Paid users get the full connector range alongside Opus or Sonnet. Claude Cowork launched earlier this year with the same connector infrastructure, and today's update brings that capability into voice for the first time.
OpenAI released GPT-Live, a new voice model for ChatGPT with improved conversational flow and more natural interruption handling. GPT-Live cannot take actions in external apps via voice. Anthropic made the opposite tradeoff: no voice model upgrade, but voice can now do things text-mode Claude can do - which is a more useful distinction for anyone who wants to talk their way through real work rather than just have a natural-sounding conversation.
11 Languages Now Supported
Claude voice mode now supports 11 languages. Switch by telling Claude which language to use - the switch can be done by voice, mid-conversation. Supported languages:
- English
- French
- German
- Hindi
- Indonesian
- Italian
- Japanese
- Korean
- Portuguese (Brazilian)
- Spanish (Latin America)
- Spanish (Spain)
One Limitation Worth Noting
Anthropic did not update the underlying voice model in this release. No changes to interruption handling, audio quality, or latency were announced. Voice mode is available in beta across mobile, desktop, and web - the Opus and Sonnet upgrade is available to all users on paid plans. Anthropic has not said when the voice model itself will receive a quality update to close the conversational gap with OpenAI.