The first version of Sora was remarkably poor at video generation, given that it comes from the best known AI company, OpenAI, creator of ChatGPT.  But Sora 2, released in the US in October 2025, made a lot of waves because like Veo 3 it can incorporate speech in the video and it has high realism.  It creates videos that are 10s long, rather 8s as for Veo 3. It is less strictly censored than Veo. It has an interesting 'cameo' feature that allows you to scan your face and then incorporate a very accurate copy of your face in your videos. you can also use cameos of other people, if they have authorized that. Here are some sample videos: