@Soli top MIT text-to-video is ranked only 14 among all.

The top tens models are proprietary

@Soli still not.

Sora 2, Google veo 2 are not open source

https://lmarena.ai/leaderboard/text-to-video

These should resolve yes.

See this leaderboard

Haven’t looked into it too much but I'm pretty sure some are open source

@ChinmayTheMathGuy December 2025, no open source in the top;

A chinease company released a sora competitor. It seems that more and more people are figuring out how this works so maybe we will see an open-sourcr alternative sooner than i thought

Comparable to current SORA or to the version that will exist then?

@0482 to current Sora

OpenAI basically hired one of the guys from the Diffusion Transformers paper and helped him scale it up. Paper came out December 19, 2022; initial model finished training March 2023. Then they kind of sat on it all year while making improvements.

Manifolders consider themselves AI experts, so some trader here will surely knock out a clone in 3 months to claim the free mana. So definitely this year.