OS 12B model that beats Claude 4 Sonnet at video captioning and is 17x cheaper

hf: https://huggingface.co/inference-net/ClipTagger-12b

blog post: https://inference.net/blog/cliptagger-12b

docs: https://docs.inference.net/use-cases/video-understanding

serverless API: https://inference.net/models/cliptagger-12b