Claude Voice Mode to get Opus and Sonnet model options

Anthropic has enabled its Opus and Sonnet models to power the voice mode in its Claude AI assistant, according to a report published Wednesday. The company previously offered a model selector in the voice interface for roughly three weeks, but the option was cosmetic because all sessions ran on the older Haiku 4.5 model regardless of the selection. That changed Wednesday when selecting Opus or Sonnet began routing voice conversations through those models instead of falling back to Haiku. The picker currently exposes Opus, Sonnet, and Haiku alongside a control for setting the thinking level. The company's Mythos-class model, Claude Fable, is absent from voice for now. The pipeline remains text-to-speech rather than a speech-native model, with spoken output appearing to rely on ElevenLabs. The report said the functional switch behind the selector suggests a formal public rollout is close, though no announcement has appeared. In testing, interruption handling held up as the model pauses when it hears the user and resumes sensibly. The report noted the direction contrasts with OpenAI's pursuit of a bidirectional, speech-native model, as Anthropic is placing its strongest reasoning models on a conventional voice stack.