Video AI models still lag roughly two years behind text-based large language models, according to Amit Jain, the CEO of Luma Labs, but the pace of improvement means what took months to achieve a ...