Anthropic announced on July 23, 2026, a major update to Claude’s voice mode, officially introducing support for the Opus and Sonnet models, along with new cross-app ecosystem automation capabilities. Users can now trigger automated actions in apps such as Gmail, Google Calendar, Slack, Canva, and Notion via voice commands. The update supports 10 major languages: English, French, German, Hindi, Indonesian, Italian, Japanese, Korean, Portuguese, and Spanish. Free users are limited to connecting one app when using the Haiku model, while paid users can unlock all models and multi-app collaboration features, marking a significant step forward in AI voice interaction—from basic Q&A to complex task execution.文章作者、來源：AIBase
Anthropic 於 2026 年 7 月 23 日宣布對其對話助手 Claude 的語音模式進行重大更新，正式引入 Opus 與 Sonnet 模型支援，並新增跨應用生態的自動化協同能力，標誌著 AI 語音交互從基礎問答向複雜任務履約邁出重要一步。
此前，Claude 的語音模式僅依賴輕量級的 Haiku 模型運行，雖具備極高的響應速度，但在處理深度推理及複雜任務時存在明顯瓶頸。本次升級後，語音模式將預設繼承用戶在文字聊天中所選用的最後一個模型（涵蓋 Opus、Sonnet 及 Haiku），並在各模型內部自動匹配最快版本以兼顧響應效率。
Meanwhile, the voice mode has officially integrated with the ecosystems of Gmail, Google Calendar, Slack, Canva, and Notion, allowing users to trigger complex actions such as updating meeting schedules, drafting emails, or creating documents via voice commands, thereby establishing a differentiated competitive advantage in tool interoperability.
在語言支援方面，於先前測試版的基礎上，新版已正式支援英語、法語、德語、印地語、印尼語、義大利語、日語、韓語、葡萄牙語（巴西）及西班牙語（拉丁美洲／西班牙）等10種主流語言，用戶可在設定中手動選擇切換。
目前，該語音模式更新已向所有平台的 Beta 測試用戶開放。其中，免費用戶在使用 Haiku 模型時僅限連接單一應用，而付費訂閱用戶可解鎖全系模型選擇與多應用協同能力。
Although Anthropic has not yet restructured the underlying speech stack (it still relies on a turn-based dialogue mechanism rather than a full-duplex architecture), by combining the general reasoning capabilities of high-level large models with external toolchains, it has significantly expanded the application boundaries of voice interaction in commercial reasoning, product discussions, and office automation.