The Batch: OpenAI's GPT-5.2, Runway's Interactive Worlds, and Disney's AI Alliance

OpenAI’s Answer to Gemini 3: GPT-5.2

TLDR: OpenAI has launched GPT-5.2, a new suite of models, just weeks after Google's Gemini 3 announcement. GPT-5.2 boasts significant improvements in reasoning and cost-efficiency.

Link: Introducing GPT-5.2

Coherent, Interactive Worlds with Runway's GWM-1

TLDR: Runway has introduced GWM-1, a family of "general world models" that can generate coherent, interactive video scenes in real time.

Link: Introducing GWM-1

Disney Teams Up With OpenAI

TLDR: Disney has entered into a 3-year exclusive agreement with OpenAI to license its characters for use in the Sora video generation app, and has taken a $1 billion stake in the company.

Link: Disney, in a First, Lets OpenAI Use Its Characters in Sora Videos

Adapting LLMs to Any Sort of Data with SEMI

TLDR: Researchers have developed a new method called Sample-Efficient Modality Integration (SEMI) that allows a large language model to process new data types, even in specialized domains, with as few as 32 examples.

Link: Sample-Efficient Modality Integration