What AI can now do·May 13, 2024·Everyday

GPT-4o answers by voice in 320 ms, conversation speed, free to all

OpenAI

320 msAverage voice response latency

GPT-4o responded to audio in as little as 232 ms (320 ms average), down from 2.8-5.4 s in the previous Voice Mode, and became the free-tier ChatGPT model; METR 50% horizon about 7 minutes.

How far it has travelled

From a demonstration to real life, in five steps. Empty steps are the point: a demonstration is not a deployment.

Demonstratedshown once, by its developer or a researcher
yes · May 13, 2024 · 232 ms fastest, 320 ms average audio response, 'similar to human response time'; prior Voice Mode averaged 2.8 s (GPT-3.5) and 5.4 s (GPT-4). METR 1.1 horizon: 7.0 min. source
Reproducedshown again by someone else
partial · Dec 31, 2024 · Real-time voice assistants from other labs followed during 2024; none was separately verified for this entry.
Accessibleavailable to the public or to customers
yes · May 13, 2024 · Text and image in ChatGPT's free tier at launch; Voice Mode alpha for Plus 'in the coming weeks'; API at half the price of GPT-4 Turbo. source
Reliableworks most of the time on a defined task
partial · Apr 29, 2025 · Default ChatGPT model for about a year (~500M weekly users); an April 2025 personality update made it 'overly flattering or agreeable' and was rolled back after relying too much on thumbs-up/down feedback. source
Real consequenceshas changed something in the world, for good or ill
yes · Apr 29, 2025 · Sycophantic update withdrawn (Apr 2025); GPT-4o is the model named in Raine v. OpenAI and the seven Nov 2025 suits alleging suicide coaching and delusion reinforcement. source

What this could change for you

A free, always-available voice that talks back like a person became a companion for hundreds of millions, including teens and adults in crisis.

Incidents it made possible

Sources

primary source, second source.

A milestone is a dated, sourced result that moved what AI can do; a stage is upgraded only when a source shows it (a reproduction, a public release, production evidence, a documented consequence). “Unknown” is the honest default.