Stepfun API: models, pricing, and specs
Compare every Stepfun model available through Felo API — prices, context windows, and capabilities in one table. Call any of them with the same API key.
- Models
- 1
- chat ×1
- Input per 1M
- $1
- Output per 1M
- $2.7
- Largest context
- 1M
StepFun's Step 5 Preview is a reasoning model with a 1M-token context, text, image, and video input, and a 600B total, 27B activated sparse mixture-of-experts architecture.
Step 5 Preview is a reasoning model from StepFun with a 1M-token context window. It accepts text, image, and video input and returns text output, supporting up to 60 images per request and short video clips. It uses a sparse mixture-of-experts architecture with 600B total parameters and 27B activated parameters per token, and offers low, medium, and high reasoning effort settings that raise internal thinking chain length and output token consumption.
The model is built for demanding, long-horizon work: agentic coding, complex software engineering, professional knowledge work, long-document, entire-codebase, and multi-PDF processing, financial domain tasks, research, and long-horizon planning. It supports tool calling for tasks that require repeated tool calls, execution, verification, and revision, and its long-context utilization suits retrieval and reasoning across large document sets. StepFun positions it for developers, research labs, and enterprises handling large document sets.
Access is API-only, so users build their own product on top of the raw model. StepFun has promised model weights for October 15, 2026, but an open-weight release has not yet been delivered, and no technical report has been released. All published benchmark results are self-reported with no independent third-party verification, and outputs are verbose, which increases output token cost. Paid API access began on September 20, 2026.
Specifications
- Input per 1M
- $1
- Output per 1M
- $2.7
- Context
- 1M
- Max output
- 64K
- Protocols
- messages