ChatGPT GPT-Live: Conversations With No Lags or Cutoffs
ChatGPT replaced the three-step voice pipeline with a single multimodal model, GPT-Live. Full-duplex: it speaks and listens simultaneously, inserts natural backchannels, and offloads complex tasks to GPT-5.5 in the background.
OpenAI has launched GPT-Live — ChatGPT's new voice engine. The old setup worked in three steps: speech was converted to text, a language model processed the text, and the response was synthesized back into voice. Each handoff added latency and risked losing context. GPT-Live replaces that pipeline with a single multimodal model that works directly with audio.
Full-Duplex and Natural Backchannels
The full-duplex architecture lets the model listen and speak at the same time — without awkward pauses when turns switch. GPT-Live adds natural reactions: it inserts "uh-huh" and "yeah" while the user is still talking. Complex requests are offloaded to GPT-5.5 in the background, keeping the conversation from stalling.
Availability and Intelligence Levels
GPT-Live is available on iOS, Android, and in the browser. Paid users (Pro, Plus, Go) get the full GPT-Live-1 with three modes: Instant, Medium, and High. The free plan is limited to GPT-Live-1 mini — Instant mode only. Live translation is already included; screen sharing is not yet available.
Source: www.engadget.com
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →▌ Related guides
Author
Evgenii Arsentev
PhD · Chief Executive Officer, digital health
Articles · Latest articles