Claude voice mode now runs Opus and Sonnet with your connected apps

Voice was Haiku-only for quick replies; now it reasons with Opus, acts through Gmail, Slack and Canva, switches models mid-call, and speaks 11 languages.

Nowline JUL 28 2:00 AM banner

Top AI stories from the last hour

Top AI stories from the last hour

Copy markdown

  • Opus and Sonnet replace Haiku-only voice

    Claude's voice mode ran only on Haiku - fast, but weak on real work. You can now pick Opus or Sonnet from the model picker, so voice handles longer reasoning like talking through a client pitch or product research instead of just quick answers.

  • Voice can now act through your apps

    Connected tools work inside voice: Gmail, Google Calendar, Slack, Canva and Notion. A spoken request becomes a real action - schedule a meeting, draft an email, turn a pitch into a Canva one-pager - fully hands-free.

  • Switch models mid-call, hop between voice and text

    You can change models mid-conversation from the picker and move seamlessly between voice and text in one thread. Start fast on Haiku, escalate to Opus when the problem gets hard, without losing context.

  • 11 languages, all platforms, in beta

    Voice now speaks 11 languages - including Japanese, Korean, Hindi, Indonesian, Brazilian Portuguese and both Spanish variants - and is rolling out to all users in beta across every platform.

  • The free-tier catch

    Free users stay on Haiku with a single connected app; the Opus/Sonnet reasoning and full tool access are paid-plan only. For voice-first workflows, that's the line between a demo and a daily driver.