Anthropic has expanded its voice mode beyond the Haiku model, making the feature available on its more powerful Opus and Sonnet systems. The update also brings voice functionality into third-party applications including Gmail, Slack, and Canva, allowing users to interact with the AI assistant directly within their everyday workflow.

When voice mode first launched last year, it was built around Haiku, a model optimized for speed and short answers. But the company quickly noticed users treating it as more than a quick-query tool. According to a blog post, people were relying on voice to work through complex business problems, demands that Haiku wasn't built to handle. Conversations stayed fast, but often shallow. With Sonnet and Opus, both designed for harder problem-solving, the assistant can deliver richer analysis and take actions like drafting a one-page pitch from a conversation or adjusting a calendar when a train runs late. Users can also switch between text and voice mid-conversation, or swap models on the fly, say shifting from a quick Haiku exchange to a deeper Opus session as an idea develops.

The company is also expanding language support. Voice mode, previously limited to English with other languages in beta, now works in French, German, Spanish, Hindi, Indonesian, Italian, Japanese, Korean, and Portuguese. A correction note: an earlier version of this story incorrectly stated the launch year as 2026; voice mode originally launched in 2025.

Source: The Verge