<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[OpenAI 推出三款实时语音 API 模型，GPT-Realtime-2 首搭 GPT-5 级推理能力]]></title><description><![CDATA[<p dir="auto">OpenAI 于 5 月 7 日发布三款面向开发者的实时语音模型，均通过 Realtime API 提供。GPT-Realtime-2 是首款具备 GPT-5 级推理能力的语音模型，可在通话过程中边推理边保持对话流畅，支持并行工具调用，在音频智能基准 Big Bench Audio 上较前代 GPT-Realtime-1.5 提升 15.2%，在多轮对话评测 Audio MultiChallenge 上提升 13.8%，按音频 token 计费（输入 32 美元 / 百万 token，输出 64 美元 / 百万 token）。GPT-Realtime-Translate 为实时翻译模型，支持逾 70 种语言输入、13 种语言输出，可与说话者同步推进，按分钟计费（0.034 美元 / 分钟）。GPT-Realtime-Whisper 则为流式语音转文字模型，可在说话人开口同时实时输出文字，按分钟计费（0.017 美元 / 分钟）。三款模型均已通过 OpenAI Playground 开放测试。</p>
<p dir="auto">OpenAI 指出，此次发布标志着实时语音从&quot;问答式交互&quot;向&quot;能实际完成任务的语音界面&quot;的跃升——模型可在对话过程中持续监听、推理、翻译、转录并调用工具。已投入测试的企业客户包括 Zillow（用于语音代理客服）和德国电信（用于跨语言客户服务），OpenAI 称两者均反馈通话成功率与合规鲁棒性明显提升。此外，三款模型内置主动内容分类器，可在检测到违规内容时中断对话，并支持欧盟数据驻留（EU Data Residency）及企业隐私承诺。</p>
<p dir="auto"><a href="https://openai.com/index/advancing-voice-intelligence-with-new-models-in-the-api/" target="_blank" rel="noopener noreferrer nofollow ugc">OpenAI</a> | <a href="https://techcrunch.com/2026/05/07/openai-launches-new-voice-intelligence-features-in-its-api/" target="_blank" rel="noopener noreferrer nofollow ugc">TechCrunch</a> | <a href="https://9to5mac.com/2026/05/07/openai-has-new-voice-models-that-reason-translate-and-transcribe-as-you-speak/" target="_blank" rel="noopener noreferrer nofollow ugc">9to5Mac</a></p>
<p dir="auto"><div class="card col-md-9 col-lg-6 position-relative link-preview p-0">



<a href="https://techcrunch.com/2026/05/07/openai-launches-new-voice-intelligence-features-in-its-api/" title="OpenAI launches new voice intelligence features in its API | TechCrunch">
<img src="https://techcrunch.com/wp-content/uploads/2025/01/GettyImages-2170386424.jpg?w=1024" class="card-img-top not-responsive" style="max-height: 15rem;" alt="Link Preview Image" onerror="this.parentElement.remove()" />
</a>



<div class="card-body">
<h5 class="card-title">
<a class="text-decoration-none" href="https://techcrunch.com/2026/05/07/openai-launches-new-voice-intelligence-features-in-its-api/">
OpenAI launches new voice intelligence features in its API | TechCrunch
</a>
</h5>
<p class="card-text line-clamp-3">The new features could be handy for customer service systems, but OpenAI says they have applications that work across a variety of other fields, including education and creator platforms.</p>
</div>
<a href="https://techcrunch.com/2026/05/07/openai-launches-new-voice-intelligence-features-in-its-api/" class="card-footer text-body-secondary small d-flex gap-2 align-items-center lh-2">



<img src="https://techcrunch.com/wp-content/uploads/2015/02/cropped-cropped-favicon-gradient.png?w=32" alt="favicon" class="not-responsive overflow-hiddden" style="max-width: 21px; max-height: 21px;" onerror="this.remove()"/>







<p class="d-inline-block text-truncate mb-0">TechCrunch <span class="text-secondary">(techcrunch.com)</span></p>
</a>
</div></p>
<p dir="auto"><div class="card col-md-9 col-lg-6 position-relative link-preview p-0">



<a href="https://9to5mac.com/2026/05/07/openai-has-new-voice-models-that-reason-translate-and-transcribe-as-you-speak/" title="OpenAI has new voice models that reason, translate, and transcribe as you speak - 9to5Mac">
<img src="https://i0.wp.com/9to5mac.com/wp-content/uploads/sites/6/2025/08/openai.webp?resize=1200%2C628&quality=82&strip=all&ssl=1" class="card-img-top not-responsive" style="max-height: 15rem;" alt="Link Preview Image" onerror="this.parentElement.remove()" />
</a>



<div class="card-body">
<h5 class="card-title">
<a class="text-decoration-none" href="https://9to5mac.com/2026/05/07/openai-has-new-voice-models-that-reason-translate-and-transcribe-as-you-speak/">
OpenAI has new voice models that reason, translate, and transcribe as you speak - 9to5Mac
</a>
</h5>
<p class="card-text line-clamp-3">OpenAI has just released three new realtime voice models that it says will “unlock a new class of voice apps...</p>
</div>
<a href="https://9to5mac.com/2026/05/07/openai-has-new-voice-models-that-reason-translate-and-transcribe-as-you-speak/" class="card-footer text-body-secondary small d-flex gap-2 align-items-center lh-2">



<img src="https://9to5mac.com/wp-content/uploads/sites/6/2019/10/cropped-cropped-mac1-1.png?w=32" alt="favicon" class="not-responsive overflow-hiddden" style="max-width: 21px; max-height: 21px;" onerror="this.remove()"/>







<p class="d-inline-block text-truncate mb-0">9to5Mac <span class="text-secondary">(9to5mac.com)</span></p>
</a>
</div></p>
]]></description><link>https://welinux.com//topic/245/openai-推出三款实时语音-api-模型-gpt-realtime-2-首搭-gpt-5-级推理能力</link><generator>RSS for Node</generator><lastBuildDate>Mon, 18 May 2026 20:30:43 GMT</lastBuildDate><atom:link href="https://welinux.com//topic/245.rss" rel="self" type="application/rss+xml"/><pubDate>Fri, 08 May 2026 03:08:55 GMT</pubDate><ttl>60</ttl></channel></rss>