HeyGen has launched Decision 1, a model that generates no text at all: it accepts text, image, video or audio input, answers up to 400 questions, and returns a probability per question that code can act on directly. It is now available through the AI/ML API platform, with a demo scoring 8 out of 8 on gesture and emotion recognition at roughly $0.0027 per run.
Modelsslug: decision-1
Decision 1
HeyGen's multimodal decision model that outputs calibrated probabilities for questions over video, audio, image and text.
Curated Intelligence
Rohan Paul (@rohanpaul_ai)T2·Tools
58 pts
OpenRouter (@OpenRouter)T1·Tools
62 pts
HeyGen's Decision 1 is now live on OpenRouter, accepting video, audio, image and text in a single request and returning a calibrated probability for each question, such as whether a selfie video matches an ID photo or is a live capture. It handles up to 400 questions per call at roughly 100 ms latency, with input pricing at 50% off through December 1.