On July 23, 2026 (U.S. time), Anthropic overhauled the voice mode of its conversational AI, Claude, extending spoken conversations beyond the lightweight Haiku model to the more capable Opus and Sonnet. The update also lets users call up connected tools such as Gmail and Google Calendar by voice alone, so simply talking to a phone can now handle tasks like checking a schedule or summarizing an email thread.
Spoken conversations reach the higher-end models
Claude's voice mode lets you hold a full spoken conversation with the AI instead of typing prompts and reading replies. Until this update, voice mode ran only on Haiku, the model tuned for speed and simplicity. With Opus and Sonnet now added, users can carry on longer and more complex conversations entirely by voice.
According to Anthropic, support for the higher-end models means Claude can handle involved discussions and step-by-step reasoning through speech. Users can even give feedback on how Claude communicates, such as asking it to be more concise. It is an update that treats voice not as simple read-aloud, but as a genuine conversational partner.
Switch models, and move between voice and text, mid-conversation
The refreshed voice mode is built around flexible switching. You can change Claude's model partway through a conversation without losing the context of what came before. You can also move between voice and text within the same chat and pick up right where you left off. The idea is that you might start a spoken query on the go, then switch to text when you want to sit down and review things carefully.
Which models are available depends on your plan, and a voice conversation begins on whichever model you last selected in text chat. In other words, the model you normally use in text carries straight over to voice.
Connected tools, triggered by voice
The headline addition is integration with outside services. Voice mode can now call directly on connected tools such as Gmail, Google Calendar, Google Docs, and Slack. You might say "tell me today's schedule" to check your calendar, or have a long email exchange summarized out loud.
The number of tools you can connect varies by plan. Free accounts can link one tool, while paid plans let you combine more. Because you can pull up information without stopping what you are doing, the feature pairs well with hands-busy situations like commuting or working.
Languages, supported environments, and limits
Voice mode's language support expanded beyond English earlier this year, with French, German, Hindi, Indonesian, Italian, Japanese, Korean, Portuguese, and Spanish available in beta. You need to select the language manually in the app. Being able to speak to it in Japanese is an easy-to-appreciate step forward for users in Japan.
Voice mode runs across iOS, Android, desktop, and the web, though Anthropic says it is designed to work best on phones. Note that voice conversations count toward standard usage limits, and the company's newest model, Fable, is not yet supported in voice mode.
The race in voice AI is heating up
The move toward operating AI by voice is gaining momentum across the industry. Around the same time, OpenAI rolled out "ChatGPT Voice" for its desktop app. It runs on a model the company calls "GPT-Live" and claims to let users direct multiple tasks at once through speech alone. Available to paid plans on macOS and Windows, it is another sign that competition over voice as the entry point to AI is intensifying.
Summary
By widening its supported models from Haiku to Opus and Sonnet, and by connecting to tools like Gmail and Calendar, Claude's voice mode has stepped up from simple read-aloud to an assistant you can operate with your voice. Its strengths are the flexibility to switch models and jump between voice and text without losing context, multilingual support including Japanese, and a design centered on phones. Voice-first AI is a focus for every major player, making this a space whose usability will be worth watching.
