OpenAI ships GPT-Live-1 in the API; voice agent developer says it matches months of his harness work
pbbakkum · x · 2026-09-11
OpenAI's dev account announced GPT-Live-1 is now available in the API, bringing ChatGPT-style natural back-and-forth voice agents that listen while they speak, working with the models and harness you choose.
Quoting the launch, developer muratcan offered a rare production perspective: after three months building duplex voice agents (thousands of lines of harness and experiments), he tried GPT-Live-1 and recognized behaviors his team spent weeks making reliable. He details why duplex voice is brutally hard — conversational timing, orchestration, acoustic turn-taking, micro-interruptions, prosody, generalized noise — and notes that beautifully crafted voice demos abound while production deployments are nearly impossible to find; his own agents' conversions keep climbing as reliability improves.
More from coding & agent
- AXL launches remote MCP server exposing 1,284 sales and marketing APIs to Claude and Codex — EdTech_Padre · 2026-10-27
- How the Claude Code Team Reviews Code When Claude Writes 90% of It — inductionheads · 2026-10-27
- AWS on agent prompt caching: explicit + implicit patterns to slash token costs — AWS ML Blog · 2026-10-27
- Reddit Thread: Keeping AI Agents Grounded in Current Project Information — Few-Ad-1358 · 2026-10-27
- He gave his AI agent a phone number and email — then hired it as social media manager — heyshrutimishra · 2026-10-27
- Firecrawl launches Alexandria company-data API and announces $75M Series B — devdigest · 2026-10-27