OpenAI has added GPT‑Live, a full‑duplex voice mode, to the ChatGPT app. The feature can listen and speak at the same time rather than waiting for turns, and it provides live speech translation with near‑zero lag: when someone speaks, the rendered translation appears or is spoken back almost instantly.
How it works
GPT‑Live uses a two‑tier approach: the front‑end audio stream continues in real time while heavier or more complex queries are routed to GPT‑5.5 in the background for processing. This setup allows continuous voice interaction on the client side while offloading computationally expensive tasks to a more capable model.
Because the capability is built into the ChatGPT application, it is available on phones and other user devices as part of the app experience rather than as a separate interpretation product.
Implications for interpretation and conference services
Simultaneous interpretation has historically remained costly and in demand because it required rare real‑time bilingual skill. By delivering near‑instantaneous, bidirectional translation on a phone, GPT‑Live addresses that scarcity factor. That changes the economics that supported high prices for human simultaneous interpreters.
This does not automatically mean all conference interpreting becomes cheap: professional interpretation includes quality control, liability, and organizational aspects that any automated solution would need to match before fully replacing human interpreters. Nevertheless, making continuous, bidirectional translation widely and easily accessible — potentially as a free or standard part of a chat app — could structurally reshape the market.
Conclusion
The rollout of GPT‑Live shows that an ability once priced for its rarity — immediate bilingual simultaneity — can now be provided as a toggle inside a chatbot. That shift has the potential to absorb work previously priced on scarcity and will raise practical and regulatory questions about how automated and human interpretation coexist.
