Context window listed by OpenRouter.
STEPFUN MODEL · VERIFIED OCTOBER 9, 2026
Step 5 Preview
Pricing
StepFun’s flagship preview model targets agentic coding, long documents and professional work. On OpenRouter, its current list price is $1 input and $2.70 output per million tokens.
A low output rate for a 1M-context agentic model—with preview risk.
Step 5 Preview is inexpensive enough to test on coding and research workloads, but “Preview” matters: price, behavior and availability can move. Evaluate accepted-task cost and pin every comparison to a provider and verification date.
01 / API PRICING
Step 5 Preview costs $1 input and $2.70 output on OpenRouter.
USD per one million tokens, verified October 9, 2026. These are OpenRouter rates, not a guarantee of identical first-party StepFun pricing.
| Token category | OpenRouter price / MTok |
|---|---|
| Input | $1.00 |
| Output | $2.70 |
| Cache read | $0.05 |
Model ID on OpenRouter: stepfun/step-5-preview. Confirm the current price and identifier in your selected provider before production use.
02 / MODEL PROFILE
Built for long-horizon agentic work.
StepFun describes Step 5 Preview as a sparse Mixture-of-Experts flagship for real-world tasks.
Maximum completion tokens listed by OpenRouter.
Total and active parameters reported for the sparse MoE architecture.
Relevant workloads
- Agentic coding across large repositories
- Extended research over long documents
- Tool-using multi-step workflows
- Professional and financial analysis
Features to evaluate
- Text, image and video input
- Function calling and structured output
- Reasoning quality versus token consumption
- Provider uptime, latency and cache behavior
03 / COST EXAMPLES
Worked Step 5 Preview API costs
OpenRouter list rates, excluding provider fees outside token billing.
10K input + 2K output: $0.0100 + $0.0054.
100K input + 10K output: $0.1000 + $0.0270.
1M input + 100K output: $1.0000 + $0.2700.
cost = input_tokens × $1 + output_tokens × $2.70Divide token counts by one million. Cache hits use the separately listed $0.05/MTok cache-read rate where supported and reported by the provider.
04 / BUYING DECISION
Test task economics, not benchmark headlines.
Step 5 Preview may fit
- You need very long context at a modest token rate
- Your workflow benefits from tools and structured output
- You can tolerate preview-model changes
- You can run an evaluation on real repositories
Choose a stable model instead
- You require long-term behavior consistency
- Your compliance process needs a mature SLA
- You cannot absorb identifier or price changes
- Your provider region or uptime needs are strict
05 / FAQ
Step 5 Preview pricing answers
How much does Step 5 Preview cost?
OpenRouter currently lists $1 input, $2.70 output and $0.05 cache read per million tokens. Prices are provider-specific and can change.
What is the Step 5 Preview model ID?
Use stepfun/step-5-preview on OpenRouter. StepFun’s first-party API may use step-5-preview; follow the documentation for the endpoint you call.
Does Step 5 Preview support a one-million-token context?
OpenRouter lists a 1,000,000-token context window and up to 64,000 completion tokens.
Is Step 5 Preview good for coding?
It is positioned for agentic work and software engineering, but suitability depends on repository navigation, tool use, patch quality, latency, retries and total tokens on your tasks.
Why does this page say “Preview” so often?
Because the release is not presented as a final stable model. Recheck prices, model identifiers and production behavior before committing a critical workload.
Method and sources. Model positioning and availability were checked on StepFun’s official platform. Token prices, context and feature fields were checked on OpenRouter on October 9, 2026. This site is independent and is not affiliated with StepFun or OpenRouter. StepFun official platform · OpenRouter model page.