Anthropic Upgrades Claude Voice Mode with Opus and Sonnet Models

Anthropic has upgraded Claude's voice mode on July 23, 2026, by integrating its advanced Opus and Sonnet models alongside the existing Haiku model. This update allows users to switch models mid-conversation and enables Claude to access connected apps like Gmail, Google Calendar, and Slack to perform tasks by voice. While OpenAI focuses on speech-native, real-time conversations with GPT-Live, Anthropic prioritizes reasoning capabilities and tool integration through a conventional turn-based voice stack.

Claude voice mode model upgrade

• ▪Users can switch between the Haiku, Sonnet, and Opus models mid-conversation during a Claude voice session using a model picker.
• ▪Anthropic upgraded Claude's voice mode on July 23, 2026, making its more powerful Opus and Sonnet models available alongside the existing Haiku model.
• ▪Claude's voice mode cannot automatically detect language switches mid-conversation, requiring users to manually select the language or state the switch out loud.
• ▪Anthropic expanded Claude's voice mode language support to include English, French, German, Hindi, Indonesian, Italian, Japanese, Korean, Portuguese, and Spanish.
• ▪Claude's voice mode previously ran exclusively on the faster but less powerful Haiku model, which struggled with complex, deep reasoning tasks.

Reasoning model routing strategy

• ▪Claude's voice mode relies on a turn-based architecture where the system listens, pauses to think, and then responds, rather than utilizing full-duplex simultaneous processing.
• ▪Anthropic's voice mode update routes existing reasoning models through a conventional, turn-based voice stack rather than introducing a new speech-native audio model.

Connected app tool access

• ▪Users can use spoken commands in Claude's voice mode to perform actions in connected apps, such as drafting emails, updating calendar slots, or creating documents.
• ▪Claude's voice mode can now access and pull context from connected third-party applications including Gmail, Google Calendar, Slack, Canva, Notion, and Google Docs.

Competitive positioning versus OpenAI

• ▪Unlike Anthropic's tool-integrated voice mode, OpenAI's updated ChatGPT voice mode cannot use external tools or connected apps to perform tasks.
• ▪OpenAI's GPT-Live system uses a bidirectional, speech-native architecture that processes speech and generates output simultaneously, unlike Claude's turn-based pipeline.

Feature availability by tier

• ▪Access to the upgraded Claude Sonnet and Opus models in voice mode is reserved exclusively for paying subscribers.
• ▪The upgraded Claude voice mode is rolling out in beta to all users across mobile, desktop, and web applications.
• ▪Free Claude accounts are restricted to using the Haiku model and are limited to a single connected application during voice mode sessions.

7 sources