Anthropic has announced a significant upgrade to Claude’s Voice Mode, making conversations smarter and more capable by allowing the feature to use its more advanced Sonnet and Opus AI models instead of relying only on the lightweight Haiku model.
Previously, voice conversations were processed exclusively by Haiku to reduce response times. While fast, it often struggled with more complex questions. With the latest update, paid users can now have voice chats powered by the same model they use for text conversations, with the option to switch between Haiku, Sonnet, and Opus at any point.
The updated Voice Mode also gains access to connected apps like Gmail and Slack, enabling Claude to reference emails, messages, and other information when users grant permission. This makes the assistant more useful for productivity and work-related tasks.
Unlike OpenAI’s real-time GPT Live experience, Claude still uses a turn-based conversation system. It listens, pauses to process your request, and then responds, meaning conversations may feel slightly less natural. It also cannot automatically detect when a user switches languages during a conversation. Instead, users must either announce the language change or select the new language in the settings.
Anthropic has expanded language support with this release, including the addition of Indonesian, and says it plans to continue improving Voice Mode throughout the year.
The upgraded Voice Mode is now rolling out in beta across Claude’s web, desktop, and mobile apps. Free users can still access Voice Mode, but conversations remain limited to the Haiku model and only one connected app at a time. Paid subscribers receive access to the more powerful models and enhanced capabilities.
