Last updated: 1 October 2026
Last verified: 1 October 2026, against the official OpenAI GPT-4o announcement (openai.com/index/hello-gpt-4o).
OpenAI has unveiled GPT-4o (“omni”) — its latest and most advanced AI model. Twice as fast and half the cost of GPT-4 Turbo, it integrates text, audio and visual inputs and outputs, responds to speech in as little as 232 milliseconds, and is coming to free users.
Quick answer
OpenAI launched GPT-4o at its Spring Update event on 13 May 2024, introduced by CTO Mira Murati. The “omni” model reasons across text, audio and visuals; it’s twice as fast and half the cost of GPT-4 Turbo, enabling free-user access. It responded to audio in 232ms (320ms average), matches GPT-4 Turbo on text, reasoning and coding, supports 50+ languages, and demos showed visual math-solving, emotional facial-expression reading, plus a refreshed UI and macOS desktop app (Windows later in 2024).
At a glance
| Fact | Details |
|---|---|
| Made by | OpenAI |
| Type | Flagship multimodal AI model |
| Launched | 13 May 2024 (Spring Update event, introduced by CTO Mira Murati) |
| Price | Half the cost of GPT-4 Turbo (API); free-user access enabled |
| Free plan | Yes — coming to free users |
| Availability | Rolling out; support for 50+ languages |
| Best for | Virtual assistants, language translation, interactive storytelling, developers |
| Our rating | Not rated — no hands-on testing performed |
In one line: GPT-4o is OpenAI’s twice-as-fast, half-the-cost multimodal flagship that talks back in 232 milliseconds.
Key takeaways
- GPT-4o (“omni”) launched on 13 May 2024: twice the speed and half the cost of GPT-4 Turbo, per CTO Mira Murati.
- It responds to audio in as little as 232 milliseconds (320ms average) — matching human conversation speed.
- GPT-4 Turbo-level text, reasoning and coding; 50+ languages; visual math-solving and emotional-intelligence demos.
- Refreshed UI plus a macOS desktop app (Windows later in 2024); free users get access. Note: this is a May 2024 launch report.
In this article
- What is GPT-4o?
- How does GPT-4o work?
- How does GPT-4o compare to GPT-4 Turbo?
- What are the pros and cons of GPT-4o?
- How much does GPT-4o cost?
- Our take: the GPT-4o launch
- Verdict: is GPT-4o worth it?
- Frequently asked questions
- Sources
What is GPT-4o?
In a groundbreaking announcement that has sent shockwaves through the tech industry, OpenAI, the pioneering artificial intelligence research company, has unveiled its latest and most advanced AI model – GPT-4o, an acronym for “omni.” This cutting-edge technology promises to revolutionize the way we interact with AI systems, offering integration of text, audio, and visual inputs and outputs, ushering in a new era of truly multimodal AI interactions.
During the highly anticipated OpenAI Spring Update event, Chief Technology Officer Mira Murati took center stage to introduce GPT-4o, highlighting its remarkable capabilities in reasoning across various modalities. “GPT-4o represents a paradigm shift in the field of artificial intelligence,” Murati declared. “It transcends the boundaries of traditional AI models by effortlessly understanding and generating content across multiple formats, including text, audio, and visuals. This remarkable feat makes the interaction with AI more natural, intuitive, and engaging than ever before.”
How does GPT-4o work?
The launch event demonstrated GPT-4o’s capabilities live. Here is what it can do, step by step:
- Step 1: Talk to it in real time — GPT-4o responded to audio inputs in as little as 232 milliseconds, with an average of 320 milliseconds, matching human conversational speed. Ideal for virtual assistants, language translation and interactive storytelling.
- Step 2: Show it things — a researcher had it read a math equation from a phone’s camera feed, then walk through the solution step by step in conversational voice mode.
- Step 3: Let it read the room — it analyzed a researcher’s facial expression, judged their emotional state, and adapted its tone and inflection, responding in a peppy voice: “Care to share the source of those good vibes? Whatever’s going on, it seems like you’re in a great mood.”
- Step 4: Build on it — GPT-4 Turbo-level text, reasoning and coding intelligence, with new benchmarks in multilingual support (50+ languages), audio processing and computer vision.
- Step 5: Access it anywhere — via the refreshed UI and the dedicated ChatGPT desktop app for macOS (Windows slated for later in 2024), with free-user access enabled by the model’s efficiency.
How does GPT-4o compare to GPT-4 Turbo?
| Feature | GPT-4o | GPT-4 Turbo |
|---|---|---|
| Speed | Twice as fast | Baseline |
| Cost | Half the cost | Baseline |
| Text, reasoning, coding | Turbo-level performance | Turbo-level performance |
| Audio | Real-time conversational speech (232ms min, 320ms avg) | Not stated in the article |
| Languages | 50+ languages, new multilingual benchmarks | Previous benchmarks |
| Free access | Coming to free users | Paid tiers |
This comparison is drawn from the article’s own head-to-head with GPT-4 Turbo, as presented by Mira Murati at the launch event (“twice the speed and half the cost of its predecessor, GPT-4 Turbo”).
What are the pros and cons of GPT-4o?
Pros
- Twice the speed and half the cost of GPT-4 Turbo, with Turbo-level text, reasoning and coding.
- Human-speed voice (232ms), visual analysis and emotional intelligence, demonstrated live.
- 50+ languages and free-user access, plus a refreshed UI and macOS desktop app.
Cons
- Windows desktop app only promised for later in 2024.
- Full rollout was still pending at publication — “the world eagerly awaits”.
- As a May 2024 launch report, the specifics are now historical.
How much does GPT-4o cost?
One of the most significant advantages of GPT-4o is its unparalleled efficiency. According to Murati, the new model is a remarkable achievement, boasting twice the speed and half the cost of its predecessor, GPT-4 Turbo. This unprecedented level of efficiency translates into significant cost savings and improved accessibility, enabling OpenAI to bring the powerful GPT-4o intelligence to its free users – a goal the company has been striving towards for many months.
| Plan | Price | What you get |
|---|---|---|
| API | Half the cost of GPT-4 Turbo | Twice the speed; multimodal text, audio and vision access |
| Free users | Free | GPT-4o intelligence coming to the free tier |
Our take: the GPT-4o launch
What the article reports: a launch-day deep dive — Murati’s introduction, the 2× speed / half-cost claim, 232ms audio responses, vision and emotion demos, and the refreshed UI plus desktop app.
What stands out: the hard numbers — 232ms minimum and 320ms average audio response, twice the speed, half the cost, 50+ languages — unusually specific for launch coverage, and consistent with OpenAI’s official announcement.
What’s missing: no independent verification, no dollar pricing, no firm rollout dates — the piece is thorough but still single-sourced to the event.
Bottom line: the most detailed of the launch reports in this series, and a solid historical record. Read it as the story of 13 May 2024, not as current product information. Note: this article was originally published on 13 May 2024.
Verdict: is GPT-4o worth it?
On the article’s launch-day terms: yes — twice as fast, half the cost, human-speed voice and free-tier access made GPT-4o an easy recommendation the day it launched. Today the verdict holds in spirit, but every specific here is historical: check OpenAI’s current model lineup for what GPT-4o (and its successors) offer now.
Frequently asked questions
OpenAI Chief Technology Officer Mira Murati introduced GPT-4o on stage at the Spring Update event on 13 May 2024.
The live demo showed GPT-4o responding to audio inputs in as little as 232 milliseconds, with an average response time of 320 milliseconds — matching human response times in conversation.
According to Murati, GPT-4o boasts twice the speed and half the cost of GPT-4 Turbo, while achieving Turbo-level performance on text, reasoning and coding intelligence.
OpenAI said the model’s efficiency gains would enable it to bring GPT-4o intelligence to its free users — a goal the company had been working towards for many months.
Sources
- Official OpenAI GPT-4o announcement — checked during this update; confirms the 13 May 2024 launch, the speed and cost claims, and the capabilities reported in the article. (The article itself contains no outbound links.)
Arva Rangwala covers AI news, AI tools, guides and prompts for OpenAIMaster — what is new, what is worth using, and how to put AI to work.