Earlier this month, OpenAI released several new AI models. The GPT‑4.1 model, aimed at developers, introduced support for a massive 1 million-token context window, along with improvements in instruction following, coding, and intelligence. The o3 and o4-mini reasoning models deliver state-of-the-art results on several AI benchmarks.
Even after releasing these new models, OpenAI has continued to update the GPT-4o model. In March, OpenAI introduced enhancements to GPT-4o to make it more intuitive, creative, and collaborative, featuring better instruction following, improved coding capabilities, and a clearer communication style.
we updated GPT-4o today! improved both intelligence and personality.
— Sam Altman (@sama) April 25, 2025
Today, OpenAI announced another update to the GPT-4o model, focusing on improvements in intelligence and personality. This enhanced version is currently available only through the ChatGPT experience and is not yet accessible to developers via the API.
OpenAI claims that the model now exhibits better “vibes,” formatting, intuition for user needs, and other qualitative enhancements. However, they are not sharing updated benchmarks for this release, as the improvements are more difficult to quantify.
Aidan McLaughlin, who currently works at OpenAI on model design and capabilities, tweeted that this GPT-4o update is the fastest OpenAI has ever shipped for the main 4o line, suggesting that the pace of releases is accelerating.
excited about this model! tickles my brain; super engaging
— Aidan McLaughlin (@aidan_mclau) April 25, 2025
if you"ve got feedback drop it here! https://t.co/F52DyPsQzm
While benchmarks measure the core capabilities of AI models, improvements to real-world aspects like "vibes" indicate OpenAI"s increasing focus on the overall user experience and interaction style.
OpenAI hasn"t shared the official change log for this updated model update yet. We"ll update this post with the details once they become available.