A discrepancy between first- and third-party benchmark results for OpenAI's o3 AI model is raising questions about the ...

Historically, each new generation of OpenAI's models has delivered incremental improvements in factual accuracy, with ...

OpenAI’s o3 model shows inflated benchmark results; real-world tests reflect performance far below initial FrontierMath ...

By OpenAI 's own testing, its newest reasoning models, o3 and o4 -mini, hallucinate significantly higher than o1.

The jump is so steep that it may be causing some to think that AI has become Skynet. According to a new EduBirdie survey, 25% ...

OpenAI's latest AI models tend to make things up — or "hallucinate" — substantially more than earlier versions.

In December 2024, OpenAI held a livestream on YouTube and other social media platforms, announcing the o3 AI model. At the time, the company highlighted the improved set of capabilities in the large ...

OpenAI’s newest AI model, o3, is at the center of a growing controversy after third-party tests revealed performance significantly lower than the ...

OpenAI's new AI models are hallucinating more than their predecessor, as per an internal testing report released by the ...

Word to the wise, be careful about the images you post on social media. OpenAI's latest AI models, released last week, have ...

OpenAI released its latest o3 and o4-mini models last week, which can "reason" through uploaded images. This means it can ...

OpenAI has launched new AI models that can guess locations from photos without GPS metadata, using visual cues like signs and ...