← All models
Step 5 Preview · StepFun

What does Step 5 Preview cost to run in a coding agent?

Step 5 Preview costs about $0.025 per average coding-agent request on our measured traffic, at list prices of $1 per million input and $2.70 per million output tokens. It scores 44 on the Artificial Analysis Intelligence Index, 13th of the 48 models we track. Source: Artificial Analysis (artificialanalysis.ai)

List prices: OpenRouter (vendor endpoint), checked 2026-10-13. Request costs: Standard Compute gateway data, 79,301 coding-agent requests from 75 users, Sep 13 to Oct 12, 2026. How we compute this

Run current Claude, GPT, Gemini and open models in your agent from $19/month

See plans

What Step 5 costs per coding-agent request

Cost at list priceStep 5
Prompt under 100K tokens50% of requests, 87% of input cached$0.012
Prompt of 100K to 272K39% of requests, 88% of input cached$0.030
Prompt over 272K tokens11% of requests, 86% of input cached$0.069
Average request$0.025
100 requests$2.48
Median session16 requests; a session ends after 10 idle minutes$0.33
A month of regular usemedian of 20 users active on 5+ days$35.48

Each logged request priced at the vendor’s list rates for its own token counts, including cache reads and any higher rate for long prompts. Average output: 698 tokens per request.

Step 5 benchmarks, speed and latency

Artificial AnalysisValueRank
Intelligence Index10 evaluations, 0 to 1004413 of 48
Terminal-Bench 4.0agentic coding in a terminal33.3%13 of 39
Output speedmedian90 tokens/s25 of 44
Time to first answer tokenincludes thinking time25 s15 of 44

Source: Artificial Analysis (artificialanalysis.ai). Measured variant: Step 5 Preview, speed on StepFun. Fetched October 13, 2026. Rank is among the models on this site, older versions included. Artificial Analysis runs Terminal-Bench 4.0 in its own setup, so its scores differ from the tbench.ai leaderboard.

What to watch for

Cost per request is not cost per finished task: a weaker model can need more steps for the same work, so test it on a task from your own repository.

Where Step 5 sits on the Artificial Analysis Intelligence Index

Artificial Analysis Intelligence Index bar chart with Step 5 Preview highlighted, from artificialanalysis.ai, October 13, 2026
Artificial Analysis Intelligence Index v4.3.2, Step 5 Preview highlighted, downloaded October 13, 2026. Source: Artificial Analysis (artificialanalysis.ai)

Leaderboards

About Step 5 Preview

Step 5 Preview is StepFun's flagship model for agent work, on OpenRouter since October 8, 2026 and served there by StepFun. It is a sparse mixture-of-experts model with 27B active and 600B total parameters, reads text, images and video, and has a 1M-token context with 64K output. Artificial Analysis scores it 44 on its Intelligence Index and 33.3% in its Terminal-Bench 4.0 run, ahead of Gemini 3.8 Flash on both, at about 90 output tokens per second. In our measurements it costs $0.025 per request, about half of Gemini 3.8 Flash. Cline added it to its free model list on October 13. It is still a preview.

Sources: OpenRouter listing · Artificial Analysis · Cline v4.1.24 release notes

Agents that support Step 5

Mid-priced agent loops that want more capability than a Flash-class model.

Step 5 Preview FAQ

How much does Step 5 Preview cost per coding-agent request?
$0.025 on average at list price, $2.48 per 100 requests, and $0.33 for a median 16-request session. That is 79,301 real coding-agent requests (Sep 13 to Oct 12, 2026) priced at StepFun's list rates; the median request sent 99,534 input tokens, 87% of input was read from cache, and requests returned 698 output tokens on average.
Does Step 5 Preview charge more for long prompts?
No. Step 5 Preview bills the same rate per token up to its 1M-token context window.
How fast is Step 5 Preview?
Artificial Analysis measures Step 5 Preview at 90 output tokens per second, with 25 seconds to the first answer token, thinking time included, on StepFun's API. Source: Artificial Analysis (artificialanalysis.ai).
Is Step 5 Preview good for AI agents?
Mid-priced agent loops that want more capability than a Flash-class model. On our scoring, which is calibrated to public benchmarks, it rates 7.6/10 for agentic ability and 8.2/10 for output quality. Scores above Gemini 3.8 Flash on Artificial Analysis (44 vs 41) and Terminal-Bench 4.0 (33.3% vs 19.7%).
What is Step 5 Preview's context window?
Step 5 Preview has a 1M-token context window, large enough to hold sizeable codebases or document sets in a single run. StepFun is the maker.
Can I self-host Step 5 Preview?
No. Step 5 Preview is a proprietary, hosted-only model from StepFun. You access it through an API, such as OpenRouter, rather than running the weights yourself.
What are the downsides of Step 5 Preview?
The main trade-offs: Still a preview, so behavior and price can change; Only StepFun serves it on OpenRouter, so there is no provider fallback. It's strongest for mid-priced agent loops that want more capability than a Flash-class model.

Step 5 compared

The agent matters as much as the model. Which agent gets the most out of Step 5? →

Current Claude, GPT, Gemini and open models are included with Standard Compute

Standard Compute

Current frontier models Claude Opus 5.5 · Claude Sonnet 5.5

Your agents never stop.
Your bill never grows.

Frontier models when it counts. Efficient models when it doesn’t. The most intelligence per dollar. One flat bill.

Plans from $19/mo · cancel anytime · 7-day fair refund

7-day fair refund No credit card needed Billing by Stripe1,800+ agent users