Anthropic has removed a major limitation from Claude's voice mode by allowing users to speak with the AI's most capable models, not just its fastest one. Previously, voice conversations only worked with Haiku, Claude's smallest and speediest model. Now users can choose Haiku, Sonnet, or Opus when speaking to Claude, opening up voice interactions for complex problem-solving tasks that previously required typing.

Why Did Voice Mode Only Work With Haiku Before?

AI companies typically use lighter, faster models for voice conversations because they enable more natural, flowing back-and-forth exchanges. The tradeoff is that smaller models don't "think" as deeply, so they stumble on complex queries. Haiku is designed for speed and efficiency, while Opus represents Anthropic's most capable model, built specifically for intricate problem-solving and reasoning tasks. By locking voice mode to Haiku, Anthropic was essentially forcing users to choose between convenience and capability.

The update addresses a real user frustration. Someone wanting to discuss a complicated coding problem, analyze a research paper, or work through a multi-step business challenge would have had to type their request rather than speak it, defeating much of the appeal of voice interaction.

How to Use Claude's Expanded Voice Mode

• Model Selection: Open the same dropdown menu you use to switch models in text chat, then select Haiku, Sonnet, or Opus before tapping the waveform icon to start voice mode.
• Flexible Conversation Flow: Start a conversation by typing, switch to speaking mid-thread, or type while speaking without losing context or having to repeat yourself.
• Connected App Actions: Use voice commands to act on connected services like Gmail, Google Calendar, Google Docs, Slack, Notion, and Canva, though Claude will ask for confirmation before executing tasks to prevent mistakes.
• Language Switching: Switch between French, German, Hindi, Indonesian, Italian, Japanese, Korean, Brazilian Portuguese, and Spanish mid-conversation by simply asking Claude to do so.

The flexibility is notable. Your conversation history carries over when you switch between typing and speaking, so you don't have to catch Claude up on context. Time spent in voice mode counts against your normal usage limits, just like text conversations.

What About Safety and Accuracy With Voice Commands?

Spoken commands carry a higher risk of misunderstanding than typed text. If you accidentally ask Claude to send an email to the wrong person or delete a calendar event, the consequences could be real. Anthropic built in a safeguard: Claude asks for confirmation before executing any action on a connected app. This confirmation step gives you a chance to catch and correct any misinterpretation before something goes wrong.

The voice mode inherits whatever access permissions you already granted in text chat, so there's no separate security setup needed. However, the full feature set remains locked behind Anthropic's paid subscription tiers. Free users can only access Haiku with a single connected app, though all supported languages are available to everyone.

What Does This Mean for Claude Users?

The update removes a significant barrier to using Claude's most powerful capabilities hands-free. Professionals who dictate notes, developers who prefer speaking through problems, and anyone multitasking while needing AI assistance now have a genuine option. The addition of nine new languages also signals Anthropic's push to make Claude's voice mode a global feature, not just an English-language convenience.

This move also reflects broader competition in the AI assistant space. As voice interaction becomes table stakes for consumer AI products, Anthropic is ensuring Claude doesn't force users to sacrifice capability for convenience. The update is live now for paid subscribers across all supported platforms.