If OpenAI is able to unseat Anthropic as the top AI model for software engineering, it could help the company take away a sizable chunk of business from one of its chief competitors. — Photo by Levart_Photographer on Unsplash
OpenAI is reportedly preparing to release GPT-5, its much-anticipated next-generation AI model, sometime in early August. The move comes after half a year of teasing. Back in February, OpenAI CEO Sam Altman posted on X that GPT-5 would unify the company’s GPT and O series of AI models. That would hopefully simplify things: GPT models, like GPT-4o and GPT-4.5, are traditional AI models, while the O series, like o3 and o4-mini, are reasoning models, meaning they can think through how to solve a problem in multiple steps. At the time, Altman said that GPT-5 would be released within months.
But in April, Altman posted that GPT-5 had been delayed by an additional few months. He said that while the model was shaping up to be a marked improvement over its predecessors, OpenAI’s team found it “harder than we thought it was going to be to smoothly integrate everything.”
Now, according to The Verge, OpenAI expects GPT-5 to release sometime early next month, alongside smaller versions of the model that will presumably be cheaper for developers to use via OpenAI’s API. The main version of GPT-5 and a mini-sized version of the model will be available in ChatGPT, while an even smaller “nano”-sized model will be exclusive to the API.
The Information reports that GPT-5 is significantly improved in several key areas, most notably in software engineering. The new model is said to not just be better at answering academic questions about code and winning coding competitions, but “also at more practical programming tasks that real-life engineers might handle, like making changes in a large, complicated codebase full of old code.” According to the Information’s sources, GPT-5 performs better than Claude Sonnet 4, Anthropic’s flagship mid-sized model.
This is a potentially huge deal for OpenAI, as the concept of building software applications in tandem with an AI assistant, also known as vibe coding, continues to explode in popularity. Replit, a popular vibe coding platform, grew from US$10mil to US$100mil (RM42.3mil to RM423.5mil) in annual recurring revenue in less than a year by betting on vibe coding, but OpenAI has largely missed the vibe coding boat, because most platforms, including Replit, rely on models produced by OpenAI rival Anthropic.
If OpenAI is able to unseat Anthropic as the top AI model for software engineering, it could help the company take away a sizable chunk of business from one of its chief competitors, and bolster its own status as the AI industry’s undisputed leader. – Inc./Tribune News Service
