Claude's voice mode can now use Anthropic's more capable models and switch between them mid-conversation, plus act inside apps like Gmail and Slack — though the underlying speech model is unchanged.
2026-07-23 · by Moe Ameen
On July 23, 2026, Anthropic upgraded Claude's voice mode so it runs on the company's more capable models. When voice mode first launched, it was powered by Haiku — Anthropic's fastest, lightest model, good for instant replies but not built for complex work. Users can now choose Opus or Sonnet, or stay on Haiku, and switch between them during a live conversation. By default, voice mode uses the fastest version of whichever model you last used in text chat.
The update also lets voice mode act inside your apps. Through connectors it can reach Gmail, Google Calendar, Slack, Canva, and Notion, so you can ask it to reschedule a meeting, draft an email, or pull something from a doc by talking. Anthropic paired the model upgrade with expanded language support and interface changes, including a now-functional in-call model selector. Voice mode is rolling out to all users in beta across Anthropic's mobile, desktop, and web apps; free users are limited to Haiku and a single connected app.
Notably, Anthropic did not change the underlying speech model itself. The upgrade puts smarter reasoning behind the voice interface, but it does not bring the conversational refinements — like better interruption handling — that OpenAI shipped in its recent voice update. In short, Claude got more capable at what it says, not necessarily smoother at how it listens and talks.
Voice mode is now good enough to be where you think out loud — but thinking out loud is not shipping. The practical move: talk through your week's content on a walk, dictating raw ideas, angles, or a rough script into Claude voice mode with Opus behind it, then bring that text into Kompozy to actually build and publish it. Paste the transcript or notes into Kompozy's Quick Ingest and one idea fans into 25–35 finished outputs — captioned Persona Shorts, carousels, quote graphics, a blog draft, a newsletter, and text posts — all held to your Persona Brief so the batch reads as your brand.
Where Claude voice mode stops at a smart conversation, Kompozy carries it to the finish. It renders each format, keeps the face and voice consistent across avatar video, and schedules everything across eight social platforms plus blog and email with autopilot and a per-post review pipeline. Voice mode helps you decide what to say; Kompozy makes it and ships it.
As of July 23, 2026, it runs on Anthropic's more capable models. You can pick Opus, Sonnet, or Haiku and switch between them during a conversation, where before it ran only on Haiku. It also connects to apps like Gmail, Google Calendar, Slack, Canva, and Notion, and added expanded language support. The underlying speech model was not changed.
It is rolling out in beta to all users, but free users are limited to the Haiku model and a single connected app. Access to Opus and Sonnet and more connectors comes with a paid Claude plan. Confirm current details on Anthropic's site.
Not directly. Anthropic upgraded the models doing the reasoning, not the speech model, so conversational aspects like interruption handling did not get the kind of update OpenAI recently shipped. The gain is in how capably Claude answers, not how smoothly it talks.
Anthropic said it is rolling out in beta across its mobile, desktop, and web apps.