Claude voice mode now runs on Opus and Sonnet
Reviewed by Helen Jones
Anthropic has expanded Claude voice mode so it can use the same models available in text chat, including Opus and Sonnet, rather than limiting spoken sessions to a lighter default. Users pick the model inside voice mode. Sessions start from the last text model used and can move to the latest generation the account allows. Voice turns count against normal message quotas, and access still follows each plan’s model limits. Help-center guidance and founder posts on 23 July 2026 confirmed the change across mobile, desktop, and web. Fable is still not available in voice.
Before this, voice was convenient but shallow for hard founder work. Spoken mode was fine for quick notes, yet complex offer logic, messy customer signals, or support diagnosis still pushed people back to text where the stronger models lived. Now the capability gap between talking and typing is mostly closed. You can stay in a walking or chores context and still get higher-quality reasoning on positioning, objections, and product decisions, without bouncing interfaces. The tradeoff is real: every spoken turn spends the same scarce quota you need for deep text work.
Analysis
Treat this as a trend to test for bottleneck work, not a reason to redesign your stack. Run one 15-minute Opus or Sonnet voice session on a live revenue problem such as a weak sales-page hook or a stuck nurture bridge, then immediately paste the transcript into a short written brief with the decision, proof points, and next draft so the session produces an asset instead of burning quota on loose talk.
Source note
Pulse published by Collab365 Spaces, reviewed by Helen Jones on . Cite as "Claude voice mode now runs on Opus and Sonnet", Collab365 Spaces. 3 sources referenced.