What Happened

OpenAI opened GPT-Live-1 to developers via API. The model handles listening and speaking in a single model, replacing the traditional three-module pipeline of speech-to-text, reasoning, and text-to-speech. The company also introduced OpenAI Presence, an enterprise workflow product for real-time voice interactions. Front-end pricing is set at $0.05 per minute, with backend inference and tool services billed separately.

Who Is Affected

Developers building voice agents, customer-service automation, and CX tools are directly affected. In early evaluations, the language-learning app Speak found that GPT-Live-1 gave learners more time to think and cut interruptions by almost 80% compared with the previous turn-based system. The impact spans domains including airline, retail, telecom, and banking, where spoken customer-service tasks were evaluated.

Market Impact

• The structural shift to a single voice model reduces latency: on Full Duplex Bench, GPT-Live-1 improves by 30 percentage points over GPT-Realtime-2.1, with largest gains in turn-taking latency and interactive behavior.
• Paired with GPT-6 Astra at medium reasoning effort, it ranks first on Tau3, grading voice-agent intelligence on end-to-end tasks.
• Developers no longer have to manually coordinate interruptions, pauses, or changes in direction—the front end handles those natively.
• Enterprise route via OpenAI Presence targets organizations wanting a complete workflow for real-time voice interactions.

What to Watch

Voice options are expanding but custom voices remain limited to customers who clear an eligibility check through a sales contact. Language availability will grow over coming months. Total cost per call depends on the backend model chosen (e.g., Luna for high-volume, Astra for reasoning), so budgeting from the $0.05/minute rate alone is incomplete. Integration with enterprise systems through Presence may drive deeper adoption in customer service use cases.

Related coverage

• OpenAI begins rolling out Astra model after warning of its advanced cyber capabilities
• Mistral AI Raises €3 Billion at €21 Billion Valuation
• Introducing Claude Fable 5.1 and Claude Mythos 5.1
• Nine open-weight AI models launched in 12 days, led by Thinking Machines' Inkling