Think through hard problems in voice mode
02:00 · July 23, 2026 · Claude Blog

Summary
Anthropic has expanded its Claude voice mode to run on the Opus and Sonnet models in addition to the lighter Haiku model previously used. The change allows users to handle extended, open-ended discussions that benefit from stronger reasoning rather than speed alone. A model picker lets participants switch between models during a single conversation, and the system defaults to the model last selected in text chat so sessions can move fluidly between voice and typed interaction.
Voice mode now supports connected tools once explicit permission is granted. Users can link new tools through the Connectors section in the mobile, desktop or web settings, after which Claude requests approval before executing any action. The interface remains turn-based: the model listens, pauses to process, and then replies, enabling iterative exploration of ideas such as rehearsing a pitch or reviewing a decision process out loud.
Multilingual support has also been broadened. Participants can switch languages mid-conversation either by speaking the request or by selecting an option in the voice settings menu. The system does not detect language automatically, so the chosen language must be set specifically for voice mode. The update is available in beta to all users across platforms, though it performs best on mobile devices, with usage counted against existing plan limits.
Why it matters
This update is highly relevant for product teams and builders as it enhances Claude's utility for complex problem-solving and workflow integration via voice. The addition of multilingual support and tool connectors provides new avenues for Dutch AI practitioners to streamline development and brainstorming processes.






