stepfun · Balanced
Step 5 Preview
Step 5 Preview is StepFun's 600B sparse MoE frontier model for production-scale agents, activating 27B parameters per token. It is built for software engineering, long-horizon tool use, research, professional knowledge work, and finance, with native text, image, and video understanding and a 1M-token context window. ⚠️ Note: This model routes through StepFun, so privacy and logging guarantees may be limited.
- Reasoning
- Tools
- Vision
- Structured output
- Video input
- Parallel tools
- Input
- Text, Images, Video
- Output
- Text
- Reasoning effort
- low · medium · high
- Discovery tier
- Balanced
Pricing
USD per 1M tokens- Input
- $1.00
- Output
- $2.70
- Cache read
- $0.05
Reference prices as published by the model provider. API example
Specifications
- Context window
- 1M
- Max output
- 65.5K
- Average output (7 days)
- 1.5K tokens
- Longer than 86% of 465 models
Side by side
Step 5 Preview compared
Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.
| Step 5 PreviewstepfunThis model | Grok 4.6xAIRemove | |
|---|---|---|
| Provider | stepfun | xAI |
| Discovery tier | Balanced | Frontier |
| Added | Sep 20, 2026 | Aug 12, 2026 |
| Limits | ||
| Context window | 1MBest | 500K |
| Max output | 65.54K | — |
| Average output | 1.5K tokens | 1.8K tokens |
| Pricing, USD per 1M tokens | ||
| Input | $1.00Best | $2.00 |
| Output | $2.70Best | $6.00 |
| Cache read | $0.05Best | $0.50 |
| Performance | ||
| Throughput | — | 117.5 tok/s |
| Time to first token | — | 3.7 s |
| Request success (24h) | 100.0% | 100.0% |
| Benchmark indexes | ||
| Intelligence Index | 43.7 | 44.3Best |
| Coding Index | — | 76.8 |
| Capabilities | ||
| Reasoning | Yes | Yes |
| Tools | Yes | Yes |
| Vision | Yes | Yes |
| Structured output | Yes | Yes |
| Video input | Yes | No |
| Parallel tools | Yes | No |
| Reasoning effort | low, medium, high | low, medium, high, xhigh |
Public evaluations
Benchmarks
Independent evaluation results, with how each score ranks among the models in the Noumen catalog.
- Step 5 Preview
- Grok 4.6
Intelligence Index
- Step 5 Preview43.7
- Grok 4.644.3
Step 5 Preview: Better than 89% of 326 scored models
Indexes
Coding Index
Composite of the coding evaluations
- Step 5 Preview: —
- Grok 4.6: 76.8
Agentic work
AutomationBench-AA
Workflow automation, penalized for guardrail violations
- Step 5 Preview: 51.0%
- Grok 4.6: 66.7%Top
AutomationBench-AA tasks completed
Workflows finished end to end without a guardrail violation
- Step 5 Preview: 17.7%
- Grok 4.6: 32.7%Top
AA-Briefcasev1.1
Agentic knowledge work, rated head to head
- Step 5 Preview: 1,423 Elo
- Grok 4.6: 1,531 EloTop
GDPval-AAv2.1
Economically valuable occupational tasks, rated head to head
- Step 5 Preview: 1,587 Elo
- Grok 4.6: 1,618 EloTop
Document reasoning
GDP.pdf
Professional reasoning over PDF documents, all-pass rate
- Step 5 Preview: 14.8%
- Grok 4.6: 17.0%Top
AA-LCRv1.1
Reasoning across long documents
- Step 5 Preview: 88.3%Top
- Grok 4.6: 80.3%
MLCR-AA
Long-context reasoning over medical records
- Step 5 Preview: 18.3%Top
- Grok 4.6: 15.6%
Reasoning and knowledge
Humanity’s Last Exam
Expert-level questions across academic fields
- Step 5 Preview: 46.5%Top
- Grok 4.6: 42.9%
GPQA Diamond
Graduate-level science questions
- Step 5 Preview: —
- Grok 4.6: 94.9%
Coding
Terminal-Benchv4.0
Practical software engineering in a terminal
- Step 5 Preview: 33.3%Top
- Grok 4.6: 21.2%
SciCode
Scientific computing in Python
- Step 5 Preview: 58.9%Top
- Grok 4.6: 56.5%
Legacy evaluations
AA-LCR (unversioned)
Long-context reasoning, earlier grading
- Step 5 Preview: 88.3%Top
- Grok 4.6: 80.3%
Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis
- Step 5 Preview
- Grok 4.6
| Leaderboard | Step 5 Preview | Grok 4.6 |
|---|---|---|
| Coding | Not in top 20 | #14 · 17.4% |
Ranks among Noumen catalog models on Use cases leaderboards. Only top-20 placements are listed.
- Step 5 Preview
- Grok 4.6
| Leaderboard | Step 5 Preview | Grok 4.6 |
|---|---|---|
| Vision Style Control · Humor | Not in top 20 | #12 · 1270.6 |
| WebDev · Image To Webdev | Not in top 20 | #11 · 1603.5 |
| Document Overall | Not in top 20 | #19 · 1452.3 |
| Document Style Control Overall | Not in top 20 | #18 · 1461.1 |
Ranks among Noumen catalog models on LMArena leaderboards. Only top-20 placements are listed.LMArena
- Step 5 Preview
- Grok 4.6
| Leaderboard | Step 5 Preview | Grok 4.6 |
|---|---|---|
| FrontierCode Main | Not in top 20 | #9 · 48.0% |
| FrontierCode Extended | Not in top 20 | #7 · 61.3% |
Ranks among Noumen catalog models on FrontierCode leaderboards. Only top-20 placements are listed.FrontierCode
Live routing data
Performance and reliability
Recent averages across requests routed for this model. Your own speed depends on prompt length, reasoning effort and load.
- Prompt cache hits
- 32.5%
- About $0.69 per 1M input tokens with caching
- Request success (24h)
- 100.0%
- Reliable
Cache hits are the share of input tokens recently served from a prompt cache across all requests. They are not a forecast of your own hit rate.
- Healthy (98% or more)
- Degraded (90–98%)
- Down (under 90%)
- No requests
Developers
Use it through the API
Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.
curl https://www.noumen.xyz/api/v1/chat/completions \
-H "Authorization: Bearer $NOUMEN_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "stepfun/step-5-preview",
"reasoning_effort": "medium",
"messages": [{ "role": "user", "content": "Explain this stack trace." }]
}'import os
from openai import OpenAI
client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
model="stepfun/step-5-preview",
messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)stepfun
Related models
Other stepfun models, starting with the same family.