Step 5 Preview
StepFunProprietaryReleased Sep 18, 2026
Step 5 Preview ranks #71 on LMArena (1441.7); official output price $2.70 per 1M tokens.
Key metrics
Models ranked nearby
Sorted by Arena score| # | Model | Arena | Intel. | Output $ | tok/s |
|---|---|---|---|---|---|
| 47 |  Kimi K2.5 | 1445.2 | 23.5 | $2.75 | – |
| 48 |  GPT-5.1 | 1442.7 | 24.7 | $10.00 | – |
| 49 |  Claude Sonnet 5 | 1442.1 | 38.2 | $10.00 | – |
| 50 | SStep 5 PreviewThis page | 1441.7 | 43.7 | $2.70 | 90 |
| 51 |  Qwen3.8 27B | 1441.1 | 33.7 | $3.00 | 47 |
| 52 | GPT-6 Astra | 1440.5 | 52.7 | $50.00 | 48 |
| 53 |  GLM-4.6 | 1439.8 | 14.9 | $2.20 | – |
Pricing
- Official API list price: $1.00 input and $2.70 output per 1M tokens (USD).
- At a typical 3:1 input-to-output ratio, the blended cost is about $1.43 per 1M tokens.
- Among 78 models with an output price, it is the #22 cheapest.
Sources & methodology
Step 5 Preview (StepFun): Arena 1441.7 (#71), Intelligence Index 43.7, $1.00 in / $2.70 out per 1M tokens. Data updated Oct 9, 2026.
- LMArena — Arena score, 95% confidence interval, votes and rank.
- Artificial Analysis — Intelligence / Coding Index, output speed and latency (medians).
- OpenRouter — Route prices, context length and listing date.
Arena scores use the Bradley-Terry model (not Elo); ± is the half-width of the 95% confidence interval. Latency figures are Artificial Analysis medians; for reasoning models, time to first token and to first answer include the thinking phase. Data syncs daily at 14:20 (UTC+8).
FAQ
What rank is Step 5 Preview right now?
Step 5 Preview is #50 on our leaderboard of 68 scored models (Arena score 1441.7); its raw LMArena rank is #71.
What does Step 5 Preview cost via API?
Input $1.00 / output $2.70 per 1M tokens (official API list price; the OpenRouter quote when no official price is available). At a typical 3:1 input-to-output ratio, the blended cost is about $1.43 per 1M tokens.
How fast is Step 5 Preview?
Artificial Analysis measures about 90 output tokens per second, 1.9 s to the first token and 24 s to the first answer token; it is a reasoning model, so latency includes thinking time.
How close is Step 5 Preview to the models ranked next to it?
It is 0.4 points behind Claude Sonnet 5 (1442.1) and 0.6 points ahead of Qwen3.8 27B (1441.1); gaps smaller than the confidence interval (±8.7) are not significant.
Is Step 5 Preview open source? Can it be self-hosted?
Step 5 Preview is a closed model, available only via official APIs or hosted services.