Anthropic has expanded Claude's voice mode so it can run on the company's more capable Opus and Sonnet models, rather than being limited to Haiku. The feature is available across Claude's mobile app, desktop app, and web chat, and users can switch models in the middle of a conversation. Because Opus and Sonnet offer stronger reasoning than Haiku, the change brings voice mode to tasks that benefit from deeper analysis.

The update broadens the use of voice interaction in Claude beyond Anthropic's fastest lightweight model. According to Anthropic's materials on Claude voice mode at claude.com/blog/think-through-hard-problems-in-voice-mode, the mode is designed for spoken conversations with the assistant and can connect with supported tools when those integrations are enabled.

Claude's voice mode supports eleven languages. It can also work with connected services such as Gmail, Google Calendar, and Slack, allowing users to complete tasks by voice, including composing and sending emails. Voice has become a competitive battleground among AI providers, each pursuing a different interaction model.

The main voice assistants from Anthropic, OpenAI, and Google use different interaction models. OpenAI's GPT-Live uses full-duplex audio, meaning it can listen and speak at the same time. Claude uses a turn-based approach that waits for the user to finish speaking before it responds. Google's Gemini Live follows a similar turn-based pattern, but is limited to smartphones.

In practical use, OpenAI and Google can feel more natural because of their speech processing. All three systems handle voice-based searches well. Claude's distinguishing feature is its tool integration: Anthropic is currently the only provider among the three that lets users compose and save an email directly from audio mode. As these providers continue iterating, the tradeoff between conversational naturalness and tool-connected productivity remains a defining distinction in the voice assistant market.