OpenAI’s Sora just released, barely making their end-of-year deadline after being announced in February. The text-to-video model is now available for Plus and Pro users at sora.com, offering image-to-video, text-to-video, and video remixing capabilities.
MKBHD reviewed it, and from his review plus the original announcement demos and recent leak, Sora is technically the best video model – but not by much. Minimax and even the open source Hunyuan are keeping pace with impressive capabilities of their own.
While European users are left out for now, Sora does bring some notable features to the table. The interface MKBHD showed allows for straightforward video generation from text descriptions or images. The quality is solid, though perhaps not the quantum leap some expected after such a long wait.
This launch adds to OpenAI’s recent string of releases, which I covered in detail in an earlier post about their “12 Days of AI” announcement (https://adam.holter.com/openai-plans-12-days-of-ai-releases-including-sora-and-new-reasoning-model/). The video generation space is getting crowded – just last month Tencent released their own 13B parameter model for free (https://adam.holter.com/tencent-just-released-a-13b-parameter-video-model-for-free/).
The bottom line? Sora is good, but the competition isn’t far behind. For those with OpenAI Plus or Pro access, it’s worth checking out. Just don’t expect it to dramatically outperform other options on the market.
