Anthropic has addressed one of the most persistent complaints about Claude’s voice interface: the model powering it. Since voice mode launched, it was locked to Haiku, Claude’s fastest but least capable model, designed for instant responses rather than deep reasoning. Users who relied on Sonnet or Opus for serious work found themselves dropping to a less capable model every time they switched to voice.
As of July 23, that ceiling is gone. Paid Claude subscribers can now use voice mode with Sonnet or Opus, matching the intelligence available in text-based conversations. Users can switch between models mid-conversation, and voice sessions automatically carry forward whatever model was last used in text.
The upgrade matters most for professional workflows that require hands-free interaction without sacrificing reasoning depth. Researchers can talk through complex problems with Opus-level analysis. Developers can dictate multi-step debugging sessions without dropping to Haiku’s lighter reasoning. The gap between voice and text has effectively closed for paying users, Haiku remains the default for free-tier users, while Pro, Max, Team, and Enterprise subscribers can choose their model.
Anthropic also expanded voice mode’s reach beyond English. The update adds support for nine additional languages: French, German, Hindi, Indonesian, Italian, Japanese, Korean, Brazilian Portuguese, and two variants of Spanish (Latin America and Spain). Users can switch between languages during a session without restarting.
Perhaps the most practical addition is connector support. Voice mode can now interact with external tools including Gmail, Google Calendar, Google Docs, Slack, Canva, and Notion. A user can ask Claude to draft an email, check a calendar slot, or summarize a Slack thread without touching a keyboard, then have Claude’s response delivered in the same voice session. Free-tier users can connect one app; paid users can connect multiple simultaneously.
The upgrade is available across Anthropic’s mobile, desktop, and web apps. Claude Fable 5, Anthropic’s most capable generally available model, is not supported in voice mode on any plan, and the company has not announced a timeline for adding it.
For paid Claude users who avoided voice mode because it felt like a step down from text, the calculus has changed. The Haiku ceiling is gone; the question is now whether voice itself is the right interface for the task.
Sources: Anthropic upgrades Claude voice mode with more powerful models (9to5Mac, July 23, 2026); Claude Voice Mode Now Supports Opus and Sonnet (AI Career Lab, July 23, 2026); Anthropic blog (Anthropic, July 23, 2026)