Amazon is developing a generative AI model, code-named Olympus, that can analyze images and video alongside text, The Information reported on Thursday, citing an individual with knowledge of the matter and two other people who spoke with Amazon about its plans.

According to the sources, the technology aims to enhance video search, enabling users to locate specific moments, such as sports highlights, using text prompts. The announcement may come during next week's Amazon Web Services' re:Invent [sic] conference.

The model's video capabilities surpass traditional recognition software by understanding complex visual dynamics, such as tracking a basketball's trajectory or detecting equipment issues in underwater inspections. The progress comes as Amazon pushes to reduce reliance on Anthropic, a key partner in its AI offerings.