Event ·Multi-source

OpenAI 发布 GPT-Live 系列全双工语音模型，对话更接近真人

OpenAI 发布 GPT-Live 系列全双工语音模型，包括 GPT-Live-1 和 GPT-Live-1 mini 两个版本。该模型可同时聆听与说话，每秒多次判断是否说话、打断或调用工具。复杂任务会委托给 GPT-5.5 系列后台执行。相比 Advanced Voice Mode，在 5-10 分钟对话中 GPT-Live-1 偏好得分更高。每周超 1.5 亿人使用 ChatGPT 语音交互。

Current read

OpenAI 这次把语音交互做得更自然了，GPT-Live 系列能边听边说，还能打断和调用工具，比以前的 Advanced Voice Mode 流畅很多。

11 sources74° AI heatLast updated 2026/7/8 22:41:35

Evidence

11 sources

Primary sourceIT之家

2026/7/8 22:41:35View source

Related source 1marktechpost

2026/7/8 18:08:21View source

Related source 2小互

2026/7/9 02:29:40View source

Related source 3@OpenAIDevs

2026/7/8 17:55:40View source

Related source 4宝玉

2026/7/9 05:10:24View source

Related source 5OpenAI Blog

2026/7/8 00:00:00View source

Related source 6Decoder

2026/7/8 18:18:55View source

Related source 7The Rundown AI

2026/7/8 17:44:21View source

Related source 8Simon Willison’s Weblog

2026/7/8 23:20:48View source

Related source 9berryxia

2026/7/9 00:36:52View source

Related source 10AI Will

2026/7/9 02:55:37View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story