StepFun releases Step 5 Preview open weights, a 600B MoE model with 1M context

StepFun has released Step 5 Preview's FP8 and BF16 weights on Hugging Face and ModelScope, a sparse MoE model with 600B total and 27B active parameters per token. It supports up to 1M tokens of context and is built for long-horizon agentic work. The FP8 version needs at least 4 B300 or 8 H200 GPUs to self-host, and the API charges $1.00 per 1M input tokens and $2.70 per 1M output tokens.