Just Now, Claude Overhauls Voice, 11 Languages, But No Chinese
Just now, both Anthropic and OpenAI announced major upgrades to their voice models.
Anthropic significantly enhanced Claude Voice. It now supports the more powerful Opus 4.8 and Sonnet 5 models (not just Haiku), allows switching between them mid-conversation, and seamlessly integrates voice and text chat contexts. Crucially, it can now use tools/connectors during voice conversations to interact with user services like Gmail, Google Calendar, and Slack. Claude Voice now supports 11 languages, but notably excludes Chinese.
OpenAI, in contrast, launched a fundamentally new architecture called GPT-Live for ChatGPT Voice. This is a full-duplex model capable of simultaneous listening and speaking, allowing for natural interruptions and real-time verbal feedback. It features a two-tier system: a low-latency front-end model for conversation flow and a backend GPT-5.5 for deep, delegated reasoning. This architecture allows complex tasks to be processed asynchronously without pausing the conversation.
OpenAI is bringing this advanced voice model to desktop, launching ChatGPT Voice for macOS and Windows. It features a global hotkey, can read active window content for context (Appshots on macOS), and can verbally command multiple Agents to work simultaneously in the background.
The key differences are clear: Claude's voice mode focuses on efficiently managing personal workflows via connected apps but operates in a strict turn-taking manner. OpenAI's GPT-Live aims for a completely natural, human-like conversational experience with interruption support, multi-tasking, and deeper desktop integration.
marsbitHace 2 días 07:51
