AA output speedVerified
88 t/s
Median output tokens/s (1k prompt, default provider)
Models / StepFun
Step 5 Preview
Step · Released 2026-09-20
StepFun's 20 Sep 2026 flagship preview. 1M context, 64K max output. Text, image, and video in; text out. Tool use, and reasoning effort low, medium, or high. The token list is in RMB. Knowledge cutoff is not on the docs.
Current StepFun flagship for long agent, coding, and multimodal work. USD list price is not published.
Composite—
Context1M
Input / 1M—
Output / 1M—
Knowledge cutoffunpublished
Output speed88 t/s
Statusflagship
Not enough families covered yet.Add more benchmark rows for Step 5 Preview to see the family profile.
Editor's note
Recorded from the model docs, the RMB pricing table, the 14 Oct cache-price notice, the 20 Sep launch post, and the public Hugging Face repo. Vendor benchmark claims stay off the ledger. Editorial depth still needs a desk pass.Benchmark placements
Per-benchmark positions
Where Step 5 Preview places on each public benchmark source that publishes a row. Rank counts every model with a latest row in the same test — not a universal quality score.
| Benchmark | Family | Rank | Score | Source | Date |
|---|---|---|---|---|---|
| AA output speed | Performance | #71/ 196 | Artificial Analysis | 2026-10-12 |
Family context
Strongest rows in each family
The three highest-scoring models with pages in each capability family. Where Step 5 Preview shows up, it's highlighted.
Reasoning
- 196
- 295
- 395
Math
- 152
- 252
- 350
Long context
- 189
- 286
- 385
Newest receipts
1 recentLast 120 days
Benchmark rows added to the public ledger for Step 5 Preview in the last 120 days. Older rows live in the full table below.
| Benchmark | Family | Score | Source | Days ago |
|---|---|---|---|---|
| AA output speed | Performance | 88 t/s | Artificial Analysis | 11 |
Family coverage
Best score per family
A high composite that hides a weak family is a trap. These bars surface the families where this model hasn't been publicly tested, and where it leads.
No family coverage yet.
Price vs. performance
Where Step 5 Preview sits on the scatter
Hover any dot for name, score, and input price. Keyboard: tab through the top twelve, or use the ranking below.
AnthropicOpenAIGoogleAlibabaMetaZhipuxAIDeepSeekMiniMaxXiaomiMoonshotOtherMeituanMistralNVIDIACohere
Full benchmark ledger
Every catalog benchmark for Step 5 Preview. Scores link to the original source; gaps mean no public row exists yet.
1 of 13 catalog benchmarks have a sourced row for Step 5 Preview.
Performance
Changelog
Version history
- Cache-hit input on the pay-as-you-go API changed. Cache-miss input and output stayed put. Earlier accounts keep the old cache-hit rate.
- Public on the StepFun API. Weights are on Hugging Face.
Sources
Where the numbers come from
- Step 5 Preview model docs· checked 2026-10-21
- StepFun API pricing· checked 2026-10-21
- StepFun cache-price notice· checked 2026-10-21
- StepFun launch post· checked 2026-10-21
- Hugging Face Step-5-Preview· checked 2026-10-21
Questions this model page answers
Every answer below is assembled from the dated fields on this page. Nothing is written separately for search.
What is Step 5 Preview?
Step 5 Preview is a StepFun model: StepFun's 20 Sep 2026 flagship preview. 1M context, 64K max output. Text, image, and video in; text out. Tool use, and reasoning effort low, medium, or high. The token list is in RMB. Knowledge cutoff is not on the docs.
What is Step 5 Preview for?
Current StepFun flagship for long agent, coding, and multimodal work. USD list price is not published.
What does Step 5 Preview cost?
StepFun pay-as-you-go list, checked 21 Oct 2026, in RMB per 1M tokens: cache-miss input 7, output 20. Cache-hit input moved from 0.35 to 0.70 at 00:30 Beijing time on 14 Oct 2026. Accounts that had already called the pay-as-you-go API before that time stay on 0.35. No USD list is published.
How large is the Step 5 Preview context window?
Step 5 Preview publishes a 1M token context window.
What is the VerdictPal status of Step 5 Preview?
Step 5 Preview carries Flagship status on the model atlas.
The Pack · Editorial newsletter
Know when the rules move.
One short email when a university changes its AI rule for theses, or a research tool changes its pricing or privacy terms. No tracking, no third-party analytics. Unsubscribe in one click.