Researchers define what counts as a world model and text-to-video generators do not
NOBODY TALKING: While everyone obsesses over Sora, researchers just redefined what actually counts as a world model—and text-to-video doesn't make the cut.
Why it matters
An international research team is establishing formal definitions and taxonomies for world models through OpenWorldLib, challenging industry assumptions about what capabilities constitute genuine world modeling. This matters because definitional clarity in AI research shapes which capabilities get funded, benchmarked, and pursued—and the exclusion of text-to-video generators signals a fundamental disagreement with the market narrative around Sora.
The key facts
9 to knowOpenWorldLib framework establishes formal definition of world models
Text-to-video generators (Sora) explicitly excluded from definition
International research collaboration to standardize fragmented landscape
Definitional disagreement suggests text-to-video and true world modeling are distinct capabilities
Published April 12, 2026 (very recent)
International research team released OpenWorldLib framework
Text-to-video generators like Sora excluded from world model definition
Addresses fragmented world model research landscape
Published April 2026
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: An international research team wants to bring order to the fragmented world model research landscape with OpenWorldLib. Text-to-video models like Sora are explicitly left out of their definition.