StepFun (阶跃星辰) is a Chinese AI lab building the Step series of multimodal foundation models (text, vision and audio), offered via its own API with per-token pricing and prompt caching.
