StepFun credits RadixArk for running Step 5 Preview production inference on SGLang

StepFun thanks RadixArk for helping power its Step 5 Preview in production with SGLang, the inference framework, and says the setup enables efficient, scalable inference for real-world workloads.

RadixArk says it helps run StepFun's first-party production API with SGLang at scale, using cache-aware routing. RadixArk also says Step 5 Preview has open weights and can be deployed with SGLang. These are the companies' own statements, and the report does not include independent performance measurements.