Claude Voice Mode to get Opus and Sonnet model options

Anthropic appears to be preparing a significant upgrade for Claude Voice Mode, with testing revealing support for Opus and Sonnet model options instead of relying solely on the lightweight Haiku model. The discovery suggests users may soon be able to choose more capable AI models for voice conversations, enabling deeper reasoning, improved coding assistance, and more complex spoken interactions.

The feature has not yet been released publicly and remains hidden behind a feature flag. However, recent changes observed in the Claude app indicate that Anthropic is moving closer to a broader rollout, reinforcing its strategy of making Claude a more capable multimodal AI assistant.

Claude Voice Mode to Support Multiple AI Models

Testing shows that Claude Voice Mode now includes a model selector allowing users to switch between different AI models for voice conversations.

The available options currently include:

• Opus
• Sonnet
• Haiku

Previously, selecting Opus or Sonnet had no functional impact, with all voice sessions continuing to run on Haiku. The latest build changes this behavior, routing conversations through the selected model instead, indicating that the functionality is now active internally.

Available Model Options

Model	Intended Use
Opus	Advanced reasoning and complex conversations
Sonnet	Balanced performance and speed
Haiku	Fast, lightweight responses

Better Voice Conversations With More Powerful Models

Moving Voice Mode from Haiku to Opus or Sonnet could substantially improve the quality of spoken interactions.

Potential benefits include:

• Better reasoning during long conversations.
• More accurate responses to complex questions.
• Improved coding and technical assistance.
• Stronger handling of multi-step tasks.
• More natural follow-up discussions.

Users who previously found Voice Mode limited by Haiku’s capabilities may notice significant improvements once higher-end models become available.

Voice Pipeline Still Uses Text-to-Speech

Although the reasoning model may change, Claude Voice Mode is still believed to rely on a traditional text-to-speech (TTS) pipeline rather than a speech-native AI model.

In the current implementation:

• Claude processes user speech.
• The selected language model generates responses.
• Spoken output is synthesized through a text-to-speech system.

This means the intelligence behind conversations improves without requiring an entirely new speech model architecture.

Voice Mode Architecture

Component	Function
Speech Recognition	Converts voice into text
Claude Model	Performs reasoning and generates responses
Text-to-Speech	Converts responses back into speech

Interruption Handling Remains Smooth

Early testing indicates that Claude Voice Mode continues to support natural conversations.

Reported improvements include:

• Immediate response when users interrupt.
• Smooth continuation after interruptions.
• Better handling of long pauses.
• More conversational interaction flow.

These capabilities make Voice Mode feel more like a real-time conversation rather than a sequence of independent voice commands.

Part of Anthropic’s Expanding AI Strategy

The Voice Mode upgrade aligns with Anthropic’s broader effort to make Claude a more capable AI assistant across chat, coding, and enterprise workflows.

Recent testing has also revealed several upcoming capabilities, including:

• Managed Projects for long-running workflows.
• Managed Agents integration.
• Expanded voice capabilities.
• Greater model selection across Claude services.

Together, these features suggest Anthropic is positioning Claude as a comprehensive AI platform capable of handling increasingly sophisticated tasks.

No Public Release Date Yet

Anthropic has not officially announced the upgraded Voice Mode or confirmed when the new model selector will become available.

Because the feature remains hidden behind an internal feature flag, its functionality could change before launch, and the rollout may occur gradually for users.

Looking Ahead

Anthropic’s upcoming Voice Mode upgrade could significantly enhance the Claude user experience by allowing conversations to run on the more capable Opus and Sonnet models instead of the lightweight Haiku model. The change promises better reasoning, richer conversations, and improved support for complex tasks while preserving the natural, interruption-friendly experience users expect from voice assistants.

Looking ahead, the addition of model selection, alongside features such as Managed Projects and Managed Agents, points to Anthropic’s broader vision of transforming Claude into a full-fledged AI assistant capable of supporting everything from everyday conversations to long-running professional workflows. As competition with OpenAI, Google, Microsoft, and xAI intensifies, richer voice capabilities are likely to become an increasingly important differentiator in the AI assistant market.