OpenAI Benchmarks GPT-Live-1 for Production Voice Agents: Task Completion, Turn-Taking, Latency and Tool Use

OpenAIDevs · x · 2026-09-11

OpenAI's developer team published benchmark results for GPT-Live-1 focused on what matters for production voice agents: task completion, back-and-forth conversation and turn-taking, response latency, and tool use. Full results are now available for developers building voice applications.

Related event: OpenAI launches GPT-Live-1 voice model on API(22 posts)→

Original post →

More from Models

Models channel →