{"slug": "claude-voice-mode-now-runs-on-anthropics-opus-and-sonnet-models", "title": "Claude Voice Mode Now Runs on Anthropic’s Opus and Sonnet Models", "summary": "Anthropic has unlocked its Claude voice mode for its Opus and Sonnet models, moving beyond the lightweight Haiku model, allowing users to conduct complex conversations and execute tasks in connected apps like Gmail and Slack via voice commands. The upgrade, announced Thursday, lets voice mode inherit the last-used text model and includes a picker for seamless switching, but Claude Fable 5 remains text-only for now. Free users are limited to Haiku and one connected app, while Sonnet and Opus require a paid subscription.", "body_md": "**July 24, 2026, (Inside AI) —** Anthropic has unlocked its Claude voice mode for its most powerful AI models, Opus and Sonnet, moving beyond the lightweight Haiku model that previously handled all spoken interactions. The change, announced Thursday, means users can now conduct complex, thinking-heavy conversations by voice, with the assistant capable of reaching into connected apps like Gmail, Google Calendar, and Slack to execute tasks.\n\nThe upgrade lets voice mode inherit whichever model was last used in text chat, and a new in-conversation picker allows seamless switching between Haiku, Sonnet, and Opus. This hands control to the user, rather than forcing everyone onto the fastest but least capable model. Anthropic framed the shift as making voice useful for real work—brainstorming product research, rehearsing a client pitch, or getting feedback on communication style—not just quick queries.\n\nFree users remain limited to Haiku and a single connected app, while Sonnet and Opus require a paid subscription. The feature supports multilingual input across ten languages, but users must explicitly name the language before speaking, as Claude cannot auto-detect switches. Notably absent from the voice model picker is Claude Fable 5, Anthropic's most capable widely available model, which stays text-only for now.\n\n## Voice That Acts, Not Just Answers\n\nThe most consequential upgrade is Claude's ability to take action in connected apps through voice commands. Users can now ask Claude to summarize their Gmail inbox, update a Notion document, or check a Slack channel—all without touching a keyboard. This pushes voice mode from a passive Q&A tool toward an active agent that completes tasks, a frontier where rivals like OpenAI and Google have already staked claims.\n\nAnthropic's approach avoids new audio infrastructure by routing existing models, which explains the rapid rollout. However, the company declined to detail its voice stack, leaving the technical underpinnings opaque. This contrasts with OpenAI's published research on its voice engine and Google's disclosures about Gemini's audio processing, raising questions about latency, privacy, and how Claude handles real-world noise.\n\n## Closing the Gap, But Questions Remain\n\nAnthropic's move catches it up to competitors, but the omission of Claude Fable 5 from voice mode is puzzling. That model excels at nuanced reasoning, yet its absence suggests either technical hurdles in real-time audio processing or a strategic decision to reserve it for text-based enterprise workflows. Anthropic offered no explanation, simply stating it \"stays out of voice for now.\"\n\nFor users, the upgrade means hands-free access to a genuinely capable assistant while driving, cooking, or thinking out loud—a tool that can now reason through complex problems rather than just fetch facts. But the lack of automatic language detection and the model tiering behind a paywall may frustrate those expecting a seamless experience. As voice becomes a primary interface for AI, the pressure is on Anthropic to match not just model intelligence, but also the polish of its rivals.", "url": "https://wpnews.pro/news/claude-voice-mode-now-runs-on-anthropics-opus-and-sonnet-models", "canonical_source": "https://insideai.news/news/ai-tools/claude-voice-mode-now-runs-on-anthropics-opus-and-sonnet-models/5237/", "published_at": "2026-07-24 13:18:39+00:00", "updated_at": "2026-07-24 19:08:28.068095+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-products", "ai-agents", "large-language-models"], "entities": ["Anthropic", "Claude", "Opus", "Sonnet", "Haiku", "Claude Fable 5", "OpenAI", "Google"], "alternates": {"html": "https://wpnews.pro/news/claude-voice-mode-now-runs-on-anthropics-opus-and-sonnet-models", "markdown": "https://wpnews.pro/news/claude-voice-mode-now-runs-on-anthropics-opus-and-sonnet-models.md", "text": "https://wpnews.pro/news/claude-voice-mode-now-runs-on-anthropics-opus-and-sonnet-models.txt", "jsonld": "https://wpnews.pro/news/claude-voice-mode-now-runs-on-anthropics-opus-and-sonnet-models.jsonld"}}