Seedance 2.0 Mini, a lighter variant of ByteDance's Seedance 2.0 video model, is now available, generating clips at roughly twice the speed and about half the cost of the standard model while holding characters, wardrobe, and visual style steady across multi-shot sequences.
Roughly twice the speed of Seedance 2.0.
About half the cost of the standard model.
Multi-shot consistency for characters, wardrobe, and visual style.
Seedance 2.0 Mini trades some of the full model's overhead for throughput. It runs at approximately twice the speed and roughly half the cost of standard Seedance 2.0, positioning it as the iteration tier in the family. Teams working a shot can run more generations for the same spend, then move heavier work to the standard model once the framing lands. The Mini variant joins the existing Seedance 2.0 options we covered at launch.
The capability that defined Seedance 2.0 stays intact in Mini: it maintains consistent characters, wardrobe, and visual style across multi-shot sequences. For narrative work, that continuity is the difference between a usable sequence and a set of clips that drift between cuts. Keeping a character on-model across shots is one of the harder problems in AI video, and Mini holds that line at a lower price point. It is a problem competitors are chasing too, including Kuaishou's Kling 3.0, which built native multi-shot video and audio into its own release.
ByteDance has been widening the Seedance 2.0 family rather than replacing it. We covered the model's move to official 4K output alongside the company's tease of Seedance 2.5, which it says can generate 30-second clips in a single pass. Mini fits that pattern, adding a tier tuned for cost and speed next to options aimed at final quality. The result is a lineup creators can sort by what a given shot needs.
NVIDIA used its SIGGRAPH announcements to line up a wave of creative software behind Model Context Protocol (MCP), the open standard that lets an AI agent read an application
NVIDIA's Synthetic Video Detector Puts a Deepfake Score Inside Newsroom Workflows July 20, 2026 NVIDIA introduced the Synthetic Video Detector NIM microservice at SIGGRAPH, adding an AI-assisted signal that flags whether a video clip contains synthetic content. It is part of
Alibaba's Qwen team has opened preview access to Qwen3.8-Max , a 2.4-trillion-parameter model the company describes as its most capable system yet, with open weig
