Claude voice mode now runs on advanced reasoning models

Claude's voice mode has quietly become something more substantial. On 23 July 2026, Anthropic announced that its advanced Opus and Sonnet reasoning models are now available through conversational speech. That is a more significant shift than it first appears.

Most people still think of voice interfaces as glorified dictation tools. This update suggests something different is happening.

What Anthropic actually changed

Until now, Claude's voice mode ran on a lighter model optimised for speed and natural conversation. The reasoning depth available in Claude Opus and Sonnet, the models most UK professionals use for complex analysis, stayed locked behind a text interface.

The upgrade changes that. Users can now access Opus and Sonnet directly through speech, and switch between models mid-conversation without losing context. The voice mode also connects to workspace tools including Slack and Google Calendar, meaning spoken commands can trigger real actions in the platforms your team already uses.

Multilingual support has expanded too, with English, French, and German among the languages now supported. The update is available in beta across mobile, desktop, and web apps.

Access is available on existing Claude plans. There is no separate enterprise tier required to use it.

Why this matters for SME leaders and operations managers

For most of the past two years, AI adoption in smaller businesses has followed a familiar pattern. A team member opens a browser tab, types a question, reads the response, and then switches back to whatever they were doing. The AI sits at a desk, not in the room.

Voice mode with reasoning capability changes that dynamic. An SME leader can now think out loud about a client proposal during a commute and receive substantive analysis back, not just a transcription of their own words. An operations manager can ask Claude to summarise the morning inbox highlights or reschedule a conflicting meeting, hands-free, while walking between sites.

This is the 80 per cent people and culture dimension of AI adoption made visible. The technology has been capable for some time. What this update does is remove the interface barrier that kept it tethered to a keyboard.

The workspace integrations are worth noting separately. Direct connections to Slack and Google Calendar mean the assistant can act, not just advise. A spoken instruction to move a meeting or flag an urgent message becomes a completed task rather than a note to action later. For operations managers handling high volumes of coordination work, that distinction matters.

At gecco, our work building custom AI assistants for SME teams consistently shows that reducing repetitive administrative work is where managers gain the most immediate time back. Voice-enabled reasoning makes that benefit available outside the desk environment for the first time.

Three things to try within five working days

1. Open Claude on your mobile device and switch the voice mode model to Sonnet or Opus. Spend ten minutes talking through a decision you are currently working on. Note whether the depth of response differs from what you have seen in text mode. This takes under thirty minutes and requires no new subscription.

2. Connect your Google Calendar to Claude and ask it to read back your schedule for the next three days, identify any conflicts, and suggest a resolution. Confirm that the calendar integration is active in your Claude account settings before testing this. The goal is to complete the task without touching a keyboard.

3. Test the Slack integration by asking Claude to summarise the three most recent messages in a specific channel and draft a brief reply. This works best in channels with high message volume where manual reading creates a time drain. Your team's existing Slack plan and Claude subscription are the only requirements.

Considerations and limitations

The update is currently in beta. That means behaviour may be inconsistent and some features may not work reliably across all devices or platforms at launch.

The workspace integrations require active connections between your Claude account and the relevant platforms. If your organisation manages those platforms through a shared IT setup, you will need administrator access or IT support to enable them.

Language support is currently confirmed for English, French, and German. Teams working in other languages should test carefully before committing workflows to voice-based interaction.

Model switching mid-conversation is available, but users should be aware that switching from a faster model to Opus may introduce a short latency increase. For time-sensitive tasks this is worth factoring in.

There are no regulatory notes specific to this update, but any voice interaction that involves client data or commercially sensitive information should be treated with the same care as text-based AI use. Speak to your data controller if you are unsure.

Where to go from here

If you lead a team and you are weighing up where voice-enabled AI reasoning fits into how your people actually work day to day, the AI Readiness survey is built precisely to surface that kind of question. Taking the survey gives you access to 65 or more free resources and a custom AI Readiness report, which leads to a free 45-minute AI Readiness call to walk through the results with you.

If you want to go further and build structured AI assistants for your team, gecco's AI Assistants service designs custom workspaces and specialist assistants for SME functions, built around how your business already operates.

Website · LinkedIn · Case Studies · Newsletter