What happened
Perplexity has plugged OpenAI's gpt-live-1 into its voice mode, Crypto Briefing reported Wednesday evening, giving the AI-search company access to the real-time audio model that currently powers ChatGPT's live voice feature. The integration lets Perplexity users talk to the assistant and get spoken responses while continuing other work, a positioning the company frames as hands-free multitasking rather than a conversational demo.
gpt-live-1 is OpenAI's production voice model and sits one layer above the older Realtime API in capability, with lower latency on turn-taking and better handling of interruptions. Perplexity has not published a technical blog post of its own at time of writing, and the terms of the arrangement, including whether Perplexity pays standard API rates or has negotiated a usage carve-out, have not been disclosed.
Why it matters
The deal is awkward on its face. Perplexity and OpenAI compete head-on in AI-powered search, with ChatGPT Search positioned directly against Perplexity's core product. Routing voice through gpt-live-1 means Perplexity's most differentiated feature now depends on infrastructure its biggest rival controls, including pricing, rate limits, and model-version access.
That's a strategic choice, not an accident. Perplexity has leaned on OpenAI, Anthropic, and Meta models in the past rather than training frontier models in-house, and the economics of a dedicated voice stack are brutal. For OpenAI, the headline is a win in a different way: it keeps gpt-live-1 at the center of the live-audio category even when the end user never opens ChatGPT.