Step 5 Preview
StepFun's flagship preview targets coding, research and professional analysis with sparse MoE architecture and a 1M-token context window.
Step 5 Preview is available on MiniRouter as . The model page has current pricing, provider routes and API setup.
About the model
- Architecture: 600B total / 27B active. StepFun's sparse Mixture-of-Experts architecture.
- Context: 1M tokens. Vercel advertises a 1M output ceiling; input and output share the context window.
- Input: Text + image. Vision input with text output; no image-generation claim.
- Release status: Preview API. Open weights are planned for October 15, not yet released as of this article.
Vercel provider pricing
Vercel provider rates checked September 21, before MiniRouter markup. Cached input is listed separately at $0.05/M tokens. These rates do not establish MiniRouter access.
| Phase | Input / 1M | Output / 1M | Blended / 1M | Effective | Source |
|---|---|---|---|---|---|
| Vercel provider rates · checked Sep 21 | $1.00 | $2.70 | $1.425 | Sep 20, 2026 – open-ended | Vercel ↗ |
Publisher-reported benchmarks
Results from the publisher’s announcement, grouped by unit.
| Benchmark | Reported result | Reported by | As of | Source |
|---|---|---|---|---|
| DeepSWE v1.1 | 67.7% | StepFun | Sep 20, 2026 | Announcement ↗ |
| Terminal-Bench 4.0 | 33.3% | StepFun | Sep 20, 2026 | Announcement ↗ |
| FrontierFinance | 66.4% | StepFun | Sep 20, 2026 | Announcement ↗ |
The announcement and Vercel listing were checked September 21; Vercel dates its release entry September 20. Publisher benchmark results are separate from MiniRouter performance.
Vercel currently advertises max_tokens, temperature and stop for this endpoint. StepFun's agent demonstrations do not establish tool-calling support through MiniRouter.
This is a preview release. The planned October 15 open-weight release is a future publisher commitment, not an available download.
Use the routable catalog to confirm MiniRouter access, supported request fields and current customer rates.
Independent measurements
Artificial Analysis evaluations of the first-party API.
Step 5 Preview ranks #11 of 462 model families on their Intelligence Index.
- Intelligence Index
- 43.7
- Composite of their evaluations
- Blended price
- $1.43
- USD per 1M tokens, 3:1 input to output, first-party list
- Output speed
- 86
- Median tokens per second
- First answer token
- 25s
- Median seconds, including reasoning
Step 5 Preview on Artificial Analysis ↗
Source: Artificial Analysis ↗, read Sep 21, 2026. Measured on the model's first-party API, not a MiniRouter route.
Keep exploring
Compare published results in the model rankings and coding benchmarks, or read more model releases.
For MiniRouter product updates, visit the changelog.
Sources
- Read at StepFun ↗
StepFun Step 5 Preview announcement
Published Sep 20, 2026
- Read at Vercel ↗
Vercel Step 5 Preview catalog
Published Sep 20, 2026
- Read at Artificial Analysis ↗
Artificial Analysis model page
Published Sep 21, 2026
Editorial reference revised 2026-09-29. Catalog status comes from the generated routing snapshot.