Frontier Models

Anthropic Enables Opus and Sonnet in Claude Voice Mode

The update routes advanced models to spoken conversations for stronger reasoning and tool integration, differing from the prior default while using a text-to-speech system.

By Marcus Vance July 23, 2026 3 MIN READ

In a contemporary open-plan technology workspace with large windows overlooking an urban skyline, an anonymous professional sits at a minimalist wooden desk facing away from the camera, their posture relaxed yet attentive while engaging in a spoken conversation with an advanced AI voice interface on a sleek silver laptop positioned centrally on the desk surface. The laptop screen displays a clean dark-mode conversational interface without any visible text, logos, or UI elements, showing a waveform animation indicating active voice input and output. Nearby on the desk rests a high-quality wireless headset with a boom microphone, a stack of technical reference books on artificial intelligence and machine learning, a notepad with handwritten diagrams illustrating reasoning chains and tool-calling workflows, and a smartphone connected via cable to the laptop suggesting seamless device integration. In the background, additional workstations feature multiple monitors showing abstract data visualizations of model performance metrics, API call logs represented as flowcharts, and graphs depicting enhanced reasoning capabilities in spoken interactions. Soft natural daylight filters through the windows illuminating neutral-toned walls adorned with subtle abstract art prints representing neural network patterns. A second anonymous figure stands near a whiteboard covered in non-text diagrams of multimodal AI architectures, holding a tablet device. The overall environment includes potted plants, ergonomic office chairs, and subtle hardware elements like external microphones and speakers positioned for optimal voice capture and text-to-speech playback. The scene emphasizes a professional, focused atmosphere dedicated to testing and utilizing upgraded AI models for complex verbal tasks, tool integrations such as web searches or code executions handled through voice commands, and differentiation from previous default model selections. Every element in the frame underscores real-world deployment of frontier AI voice capabilities in a collaborative tech setting, with attention to precise object placement including charging cables, coffee mugs, and peripheral devices that support extended spoken dialogue sessions without any graphical overlays or identifiers. — Illustration: AI Intel Report

Claude Voice Mode is Anthropic's interface for conducting spoken conversations with its Claude AI models.

Anthropic has updated its Claude Voice Mode to include support for the Opus and Sonnet models.

Previously the feature defaulted to the Haiku model.

What does the model selector change mean for Claude users?

A model selector in the Voice Mode UI had been hidden or non-functional for roughly three weeks.

It always fell back to Haiku 4.5 during that period.

Now users can select Opus or Sonnet for their voice conversations.

How do Opus and Sonnet perform in voice interactions compared to Haiku?

Opus and Sonnet enable stronger interruption handling.

The models pause on user speech and tolerate long silences.

They also support tool use via Connectors in voice conversations.

What technical approach does Claude Voice Mode employ?

Voice Mode uses a conventional text-to-speech pipeline.

ElevenLabs reportedly powers the speech synthesis in the background.

It is not a speech-native model.

Claude models and their Voice Mode capabilities
Model	Availability in Voice Mode	Interruption Handling	Tool Use Support
Opus	Selectable	Strong	Enabled via Connectors
Sonnet	Selectable	Strong	Enabled via Connectors
Haiku	Previous default	Basic	Limited
Fable	Spotted in iOS build	Not specified	Not specified

What are the implications for the AI market and stakeholders?

Rivals such as OpenAI pursue native voice models.

Anthropic's approach leverages existing frontier models in voice settings.

This could affect how users engage with AI in spoken formats.

Enhanced reasoning becomes available in voice conversations.
Better management of conversational flow through interruptions.
Integration of tools expands voice mode utility.

How have observers reacted to the Voice Mode update?

ANTHROPIC 🔥: Claude Opus and Sonnet models will soon become available on Claude Voice Mode! The new model selector has been hidden in the UI for roughly three weeks, defaulting to Haiku regardless of the selection. Today, it changed, and now it responds via Opus and Sonnet models. The system is still built on TTS; however, it handles interruptions super well and can operate Connectors!TestingCatalog, AI News Account

What developments might follow for Claude Voice Mode?

Additional languages and features like push-to-talk were spotted in iOS builds.

The model selector options include Fable as well.

Further enhancements to voice capabilities are anticipated.

Frequently asked

How can users access Opus and Sonnet in Claude Voice Mode?

Users select the desired model in the Voice Mode UI once the selector is active. The feature routes conversations through the chosen model for improved performance.

Sources

TestingCatalog — Anthropic updates Claude’s voice mode with support for Opus and Sonnet models in place of Haiku. Selecting Opus or Sonnet now routes voice conversations through those models rather than quietly falling back to Haiku. In testing, interruption handling held up: the model pauses the moment it hears you and resumes sensibly, and it tolerated long silences without cutting in early. The pipeline stays text-to-speech rather than a speech-native model. ElevenLabs in the background
pasqualepillitteri.it — It is an option, spotted in the iOS build, that would let you choose which model orchestrates voice mode among Opus, Sonnet, Haiku and Fable. Claude Voice Mode: 18 Languages, Push-to-Talk and a Model Selector Spotted in iOS Build
X — ANTHROPIC 🔥: Claude Opus and Sonnet models will soon become available on Claude Voice Mode! The new model selector has been hidden in the UI for roughly three weeks, defaulting to Haiku regardless of the selection…