I’ve been coordinating with the ByteDance team and just got Seedance 2.0 running via their official cloud services (Volcengine). To be honest, the results are stunning—the physics and temporal consistency are a significant leap over previous models.

I built seedance.fast as a streamlined interface to expose these capabilities. The model is about to be officially released, but the site is already integrated and ready for the traffic. I'm focusing on low-latency inference to make the "text-to-video" workflow actually usable for creators.

Happy to answer questions about the model's performance compared to the private betas of other leading video AIs.

I’ll be continuously streaming the Seedance 2.0 videos I generate onto my site as a live video feed. You’re welcome to follow along and get early access.

I’m also offering an option to pre-purchase and lock in access to Seedance 2.0 now. Since the operating costs for 2.0 are expected to be significantly higher, pricing will increase later. Early supporters can secure access to Seedance 2.0 at the current Seedance 1.0 pricing.

If you’re interested in experimenting early and locking in a lower rate before the price adjustment, you can join now.

website:https://seedance.fast