OpenAI launches GPT-4o, one model for text, vision and audio

OpenAI GPT-4o at a glance

GPT-4o matched GPT-4 Turbo on English text and code, responded to audio in about 320 milliseconds on average, came to ChatGPT's free tier, and cost half as much as GPT-4 Turbo in the API.

What changed with OpenAI GPT-4o

OpenAI announced GPT-4o on May 13, 2024. The o stands for omni: a single model trained end to end on text, vision and audio. OpenAI said it could answer spoken input in as little as 232 milliseconds and about 320 milliseconds on average, compared with multi-second delays in the old voice pipeline, and that it matched GPT-4 Turbo on English text and code while improving on other languages, vision and audio.

Text and image features began rolling out in ChatGPT, including the free tier, and in the API as a text and vision model that was twice as fast, half the price and had five times higher rate limits than GPT-4 Turbo. The new voice mode and audio in the API were scheduled to follow over the coming weeks for limited groups.

OpenAI GPT-4o in numbers

Values are copied from the linked source on the date shown. They are not IA Compa measurements.

Average audio response timeas of May 13, 2024 · Source for Average audio response time: openai.com (opens in a new tab): ~320 ms

Where to compare OpenAI GPT-4o

Independent leaderboards and official resources for comparing this model yourself. Rankings change often; each site shows its own current data.

Leaderboards

Artificial Analysisartificialanalysis.ai (opens in a new tab)

General leaderboards

Official resources

Official announcementOpenAI (opens in a new tab)

Why OpenAI GPT-4o matters for businesses

For businesses, GPT-4o made top-tier quality cheaper and faster, and pointed toward natural voice agents for phone support and reception.

Sources for this story

Hello GPT-4o (opens in a new tab)
OpenAI · openai.com
Official
Hello GPT-4o (archived May 13, 2024) (opens in a new tab)
OpenAI via Internet Archive · web.archive.org
Official
GPT-4o (May) model page (opens in a new tab)
Artificial Analysis · artificialanalysis.ai
Benchmark

Last verified: November 5, 2026. IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.

#llm
#openai
#gpt-4o
#multimodal
#voice-ai

OpenAI releases GPT-4, a multimodal model with human-level scores on professional exams

GPT-4 accepted text and image inputs, scored around the top 10% on a simulated bar exam, and launched in ChatGPT Plus and through an API waitlist.

Image: OpenAI

The words GPT-6 Cyber below a glowing orange horizon, with small code symbols floating in a dark sky.

Nov 5, 2026OpenAIModelsMajor

OpenAI introduces GPT-6 Cyber, a vetted-access security model rated Critical for cyber capability

GPT-6 Cyber, a security-tuned GPT-6 Astra variant, scores 48.5% on ExploitGym and found 81 high-severity Firefox flaws. As of November 5, 2026, only approved Daybreak Red customers can use it, via the API and Codex, at $20/$100 per million tokens.