Video Generation API (OpenAI-Compatible)#

Supported NIMs#

NIM

Model Name

Notes

WAN2.2-T2V-A14B-14B

wan-ai/wan2.2

Text-to-video generation (server started with NIM_MODEL_VARIANT=t2v)

WAN2.2-I2V-A14B-14B

wan-ai/wan2.2

Image-to-video generation (server started with NIM_MODEL_VARIANT=i2v)

The model field accepts only the canonical name wan-ai/wan2.2 for both variants. The variant, text-to-video or image-to-video, is fixed by the NIM_MODEL_VARIANT setting on the server, not by the model name.