Skip to content

ByteDance

Lack of transparency

ByteDance does not disclose detailed information about model architecture and inference infrastructure. Thus, the environmental impacts are estimated with a very low precision.

This guide focuses on estimating video generation impacts for ByteDance models with the video_impacts function.

Video Generation

Supported ByteDance video models:

Model Identifier
Seedance 1.0 bytedance/seedance-1.0
Seedance 1.5 Pro bytedance/seedance-1.5-pro
from ecologits.estimations import video_impacts

model = "bytedance/seedance-1.5-pro"
resolution = "1080p"
duration = 5

impacts = video_impacts(
    model_name=model,
    resolution=resolution,
    duration=duration,
)

print(impacts.energy.value)
print(impacts.gwp.value)

Use the same model, resolution, and duration values that you send to your video generation provider. You can then serialize the impacts object for logs, traces, or storage:

impact_payload = impacts.model_dump(mode="json")