CONFIRMEDDate confidence: CONFIRMEDStated as firmHow firmly the source words the date: confirmed, scheduled, estimated, delayed or unverified.Show all CONFIRMED pins89% CONFIDENCEOverall confidence: 89%How well the pin's source and references back up its dates.Weighted average of how firmly 4 references, the source included, support the pin's start and end times; a reference counts half as much for every 180 days older than the newestShow all pins at 75% confidence or better
OpenAI's[1] launch post for GPT-4o, dated May 13, 2024 and linked from its model release notes.- Start
- May 13, 202490% CONFIDENCEsource
GPT-4o Released
OpenAI releases GPT-4o, its flagship model that reasons across audio, vision and text in real time, and makes it the new default in ChatGPT.
- OpenAI announced GPT-4o ("omni") on May 13, 2024, a flagship model that can reason across audio, vision and text in real time[1][3]
- Rollout: text and image in ChatGPT that day, including the free tier, with Plus getting up to 5x higher message limits; new Voice Mode in alpha for Plus in the coming weeks; audio in the API first to a small group of trusted partners[1][3]
- TechCrunch adds a refreshed web UI, a macOS desktop app (Windows later in the year) and free-tier access to the GPT Store, memory and web search[3]
- Safety: evaluated under the Preparedness Framework with no category above Medium risk, and 70+ external red-team experts[1]
- An Apple Podcasts news episode of May 13, 2024 covers the announcement in a segment, but lists only titles[2]
Notable features
- OpenAI trained a single new model end-to-end across text, vision and audio, so all inputs and outputs are processed by the same neural network; the earlier Voice Mode chained three separate models and could not directly observe tone, multiple speakers or background noise[4]
- It takes any mix of text, audio, image and video and generates text, audio and image; audio responses take as little as 232 ms (320 ms on average), versus 2.8 s and 5.4 s for the previous Voice Mode pipelines[1]
- It matches GPT-4 Turbo on English text and code and is better on non-English text, vision and audio, and in the API is twice as fast and half the price with 5x higher rate limits[1][3]
- OpenAI reports state-of-the-art results on visual perception benchmarks, with new high watermarks on multilingual, audio and vision capabilities[4]
- A new tokenizer uses fewer tokens across 20 languages (for example Gujarati 4.4x fewer)[1]
Buy onChatGPT