Particle.news
Download on the App Store

Anthropic Adds Opus and Sonnet to Claude's Voice Mode

The beta rollout lets voice use more capable text models to perform deeper work that triggers actions in third-party apps.

Overview

  • Anthropic updated Claude’s voice mode on Thursday, July 23, 2026, to let users pick between the Haiku, Opus, and Sonnet models for spoken interactions.
  • By default voice now uses the last text model a user selected and runs its fastest variant, and users can switch models during an active voice session.
  • Voice mode can connect to third-party apps such as Gmail, Google Calendar, Slack, Canva, and Notion, allowing voice-driven actions like drafting emails and moving calendar events; free accounts are limited to Haiku and one connected app.
  • The feature set includes support for 11 languages but requires users to manually select the language they want to use.
  • Anthropic said it did not change or disclose its underlying voice stack with this release, so low-level behaviors like interruption handling, latency, and privacy trade-offs may remain unchanged and should be monitored as voice shifts toward workflow use.