Step-5-Preview
🔥 Step-5-Preview is now available!
Step-5-Preview is our flagship foundation model for real-world agentic work — a 600B-parameter sparse Mixture-of-Experts model with 27B active parameters, a 1M-token context window, and native support for text, image, and video inputs.
Weights are available on Hugging Face (
SHSLab/Step-5-Preview-BF16). Try it via our API, or deploy locally with vLLM / SGLang.
📖 Table of Contents
- Introduction
- Key Features
- Model Architecture
- Model Specifications
- Training Data
- Benchmark Results
- Agentic Capabilities
- Real-World Use Cases
- Quickstart
- Deployment
- Evaluation
- Limitations
- Ethical Considerations
- Hardware Requirements
- Performance Metrics
- Citation
- License
- Contact
🚀 Introduction
Step-5-Preview is StepFun's flagship foundation model, designed from the ground up for real-world agentic tasks. It targets professional domains such as AI coding, software engineering, professional knowledge work, and financial analysis.
StepFun's core philosophy for Step 5 is the "Pareto Frontier" — achieving the optimal balance between intelligence and cost. While previous scaling efforts focused on trading more compute for stronger intelligence, the next phase requires improving the efficiency of converting compute into intelligence.
💡 Why Step 5 Preview?
• 600B total parameters, only 27B active — near-frontier performance at a fraction of the compute. • 1M-token context window without proportional cost increases. • Competitive benchmark scores against models with 3–5× more parameters. • Built for agents — long-horizon reasoning, tool use, and autonomous execution.
Step-5-Preview represents a generational leap, with StepFun skipping the entire Step 4.x line entirely, going directly from Step-3.7-Flash to Step 5. This decision reflects the magnitude of improvement achieved in this release.
✨ Key Features
- Sparse Mixture-of-Experts (MoE): 600B total parameters, 27B active per token (~4.5% sparsity).
- 1M-Token Context Window: Equivalent to ~1,500 A4 pages, enabled by Sparse GQA.
- Multimodal Input: Text, image, and video (MP4, QuickTime, Matroska; ≤128 MB; ≤5 min recommended).
- Configurable Reasoning Effort:
low,medium,high/xhigh. - Parallel Tool Calling: Natively supported for agentic workflows.
- Strict JSON Schema Output: Reliable integration into structured systems.
- OpenAI-Compatible API: Available via Step API and third-party gateways.
- Open Weights: BF16 checkpoint available now under
SHSLab/Step-5-Preview-BF16.
🏗️ Model Architecture
92-Layer "Narrow but Deep" Design
Step-5-Preview uses a 92-layer Transformer with a narrow-deep configuration. This design is specifically intended to create longer information propagation paths for implicit multi-hop reasoning during long prefill operations.
Sparse Grouped-Query Attention (GQA) with Block-Wise Token Merging
To handle the 1M-token context window efficiently, Step-5-Preview introduces Sparse GQA with block-wise token merging. This mechanism uses sparse indexing to select only historical information relevant to the current task, reducing the number of tokens that actually enter attention computation. StepFun states this cuts indexer and top-k selection costs to approximately one-eighth of a denser baseline.
⚡ Efficiency-First Scaling
Step 5 Preview achieves near-frontier performance with 600B total parameters but only 27B active per token. This is the core of StepFun's efficiency-first philosophy.
Multimodal Encoder
The model incorporates a unified multimodal encoder that processes text, images, and video frames into a shared latent space. Video is sampled at adaptive frame rates and encoded with temporal attention, allowing the model to understand motion and long-range dependencies in screen recordings, demonstrations, and real-world footage.
📋 Model Specifications
| Category | Specification |
|---|---|
| Model Name | Step-5-Preview |
| Developer | StepFun |
| Architecture | Sparse Mixture-of-Experts (MoE) |
| Total Parameters | 600B |
| Active Parameters | 27B per token (~4.5% sparsity) |
| Layers | 92 (narrow-deep Transformer) |
| Context Window | 1,000,000 tokens |
| Attention | Sparse GQA with block-wise token merging |
| Input Modalities | Text, Image, Video |
| Output Modalities | Text |
| Video Formats | MP4, QuickTime, Matroska (≤128 MB, ≤5 min recommended) |
| Reasoning Effort | low / medium / high (xhigh) |
| Tool Calling | Parallel, strict JSON schema |
| Intelligence Index | 44 (Artificial Analysis v4.3.2) |
| Open Weights | BF16 checkpoint available now |
| API Availability | Immediate (OpenAI-compatible) |
| License | StepFun Community License |
📚 Training Data
Step-5-Preview was trained on a massive, carefully curated corpus spanning:
- Code repositories from multiple languages (Python, C++, Rust, JavaScript, Go, etc.)
- Technical documentation, API references, and software engineering forums
- Scientific papers in computer science, mathematics, physics, and finance
- Financial reports, earnings calls, and market analyses
- Multimodal data including screenshots, UI mockups, video tutorials, and screen recordings
- Agentic trajectories from simulated and real tool-use environments
The data mixture was optimized for long-horizon reasoning and tool use, with a strong emphasis on real-world professional tasks. All data was filtered for quality, safety, and license compliance. The training process used a combination of next-token prediction and reinforcement learning from human feedback (RLHF) with a focus on agentic objectives.
📊 Benchmark Results
Artificial Analysis Intelligence Index
Overall Score: 44 (Intelligence Index v4.3.2, recalibrated September 7, 2026)
The index covers 10 evaluations including AA-Briefcase, GDPval-AA v2, Terminal-Bench 4.0, SciCode, and Humanity's Last Exam.
📐 How to read these tables
- Step-5-Preview is always the first data column, for fast scanning.
- Bold marks the best score in the row across all six models.
- 🥇 flags the category leader for that benchmark.
- — indicates the model was not evaluated or did not report a score.
- All Step-5-Preview results use the
highreasoning-effort setting unless noted.
Comparison set: Step-5-Preview · GPT-6 Astra (Max) · Fable 5.1 · Claude Opus 5 (Max) · Kimi K3 (Max) · GLM-5.3 (Max)
🏷️ Open-weight vs. closed-source in this comparison
Open-weight: Step-5-Preview · Kimi K3 (Max) · Qwen3.8 Max · GLM-5.3 (Max)
Closed-source: GPT-6 Astra (Max) · Claude Opus 5 (Max)
Not disclosed: Fable 5.1
🧠 Reasoning & Knowledge
| Benchmark | Step-5-Preview | GPT-6 Astra | Fable 5.1 | Claude Opus 5 | Kimi K3 | GLM-5.3 |
|---|---|---|---|---|---|---|
| GPQA Diamond | 93.5% | 96.1% 🥇 | 93.7% | 93.2% | 93.5% | 91.7% |
| Humanity's Last Exam (HLE) | 46.5% | 54.7% | 59.1% 🥇 | 54.9% | 46.9% | 42.3% |
| AA-LCR v1.1 | 88.3% | 80.7% | 85.3% | 79.3% | 88.7% 🥇 | 79.7% |
| CritPt | 20.9% | 31.7% 🥇 | 29.7% | 29.1% | 23.4% | 19.1% |
On AA-LCR v1.1 — long-context reasoning — Step-5-Preview scores 88.3%, effectively tied with Kimi K3 (88.7%) and ahead of GPT-6 Astra by +7.6 and Claude Opus 5 by +9.0, despite those models being far larger.
📝 Benchmark descriptions
- GPQA Diamond — 198 PhD-level science MCQs (biology, physics, chemistry). Human expert average: 81%.
- HLE (Humanity's Last Exam) — Frontier academic knowledge benchmark; HLE w/ tools reported separately under Agents.
- AA-LCR v1.1 — Artificial Analysis Long-Context Reasoning.
- CritPt — 71 unpublished research-level physics challenges simulating full-scale junior-PhD research projects.
💻 Coding & Software Engineering
| Benchmark | Step-5-Preview | GPT-6 Astra | Fable 5.1 | Claude Opus 5 | Kimi K3 | GLM-5.3 |
|---|---|---|---|---|---|---|
| DeepSWE v1.1 | 67.7% | 74.1% 🥇 | 67.4% | 74.0% | 67.5% | 66.9% |
| Terminal-Bench 2.1 | 85.0% | 88.4% | 91.4% 🥇 | 89.1% | 85.0% | 83.9% |
| Terminal-Bench 4 | 33.3% | 57.9% 🥇 | 55.8% | 52.3% | 12.6% | 41.9% |
| CyberGym | 84.7% 🥇 | — | — | — | 80.0% | 84.5% |
| SciCode | 58.9% | 56.5% | 63.1% 🥇 | 56.4% | 59.5% | 59.0% |
| RoadmapBench | 54.3% | — | — | 68.3% 🥇 | 55.4% | 54.1% |
| ProgramBench | 80.5% | 85.4% 🥇 | 82.7% | 82.3% | 77.8% | 72.0% |
| SWE-Marathon v1.1 | 72.7% | 77.3% | 80.2% | 85.6% 🥇 | 84.4% | 67.4% |
| MLS-Bench-Lite | 40.5% | — | 50.3% 🥇 | 49.8% | 48.3% | 37.3% |
| SWE-Atlas-QnA | 63.6% | 60.9% | — | 66.0% 🥇 | 59.5% | 59.6% |
| SWE-Atlas-Test-writing | 50.8% | 51.1% | — | 60.3% 🥇 | 50.4% | 50.4% |
| StepCodeBench | 49.0% | 61.0% | — | 63.9% 🥇 | 43.9% | 40.2% |
| StepCode-Bench-Daily | 64.9% | — | — | 77.6% 🥇 | 57.7% | 69.1% |
| StepCode-Bench-General | 65.0% | 64.3% | — | 68.3% 🥇 | 65.2% | 62.0% |
Step-5-Preview leads on DeepSWE v1.1, StepCodeBench, ProgramBench, and SWE-Atlas-QnA among the four open-weight models in the set, and posts the best CyberGym score overall (84.7%). On Terminal-Bench 4 it reaches 33.3% — 2.6× Kimi K3 (12.6%).
📝 Benchmark descriptions
- DeepSWE v1.1 — SWE-agent harness,
temperature=1.0,top_p=0.95. - Terminal-Bench 2.1 — Verified refresh of TB 2.0; 89 curated terminal tasks across SWE, sysadmin, data processing.
- Terminal-Bench 4 — Next-generation terminal agent benchmark.
- CyberGym — Security-focused agent benchmark.
- SciCode — Scientific coding benchmark.
- RoadmapBench — 115 long-horizon coding tasks across 17 repos, 5 languages; median ~3,700 LOC changed.
- ProgramBench — Rebuild programs from binary + docs; 200 tasks, 248K+ behavioral tests.
- SWE-Marathon v1.1 — Ultra-long-horizon SWE; 20 realistic multi-hour tasks with hidden/adversarial tests.
- MLS-Bench-Lite — 30-task subset of MLS-Bench; tests inventing generalizable ML methods across 12 research domains.
- SWE-Atlas-QnA — 124 tasks on deep code comprehension across 11 production repos (Go, Python, C, TypeScript).
- SWE-Atlas-Test-writing — 90 tasks on writing production-grade unit, integration, and acceptance tests.
- StepCodeBench — StepFun internal coding benchmark; 49.0% avg@4.
- StepCode-Bench-Daily / General — StepFun internal; 553 repos, 9 task types, 20 domains, 33 languages.
🤖 Agents, Tool Use & Automation
| Benchmark | Step-5-Preview | GPT-6 Astra | Fable 5.1 | Claude Opus 5 | Kimi K3 | GLM-5.3 |
|---|---|---|---|---|---|---|
| GDPval-AA v2 | 1571 | 1580 | 1724 | 1735 🥇 | 1548 | 1634 |
| τ³-Banking | 42.5% | 41.4% | 47.2% | 42.1% | 46.0% | 50.3% 🥇 |
| AutomationBench-AA | 51.0% | 68.5% 🥇 | 59.4% | 56.6% | 58.3% | 62.2% |
| AutomationBench Public | 44.0% | — | — | — | 46.7% | 48.2% 🥇 |
| AA-Briefcase | 1417 | 1562 | 1662 🥇 | 1645 | 1492 | 1511 |
| Toolathlon-Verified | 74.1% | — | 77.8% | 80.6% 🥇 | 76.5% | 73.0% |
| MCP-Atlas | 85.6% | — | — | 87.0% 🥇 | 85.3% | 86.8% |
| PresentBench | 76.8% | — | — | 77.3% 🥇 | 75.6% | 74.5% |
| JobBench | 59.0% | — | — | 65.7% 🥇 | 54.3% | 61.4% |
| Apex-Agents | 37.8% | — | — | 41.8% 🥇 | 41.0% | 38.1% |
| Draco | 83.3% | 76.8% | 87.7% 🥇 | 87.6% | 78.5% | 82.3% |
| BrowseComp | 88.7% | 91.5% 🥇 | — | 90.2% | 91.2% | — |
| HLE w/ tools | 59.4% | 57.2% | 65.0% 🥇 | 63.6% | 56.0% | 62.5% |
Step-5-Preview outperforms GPT-6 Astra on τ³-Banking (42.5% vs. 41.4%) and Draco (83.3% vs. 76.8%), and edges Kimi K3 on MCP-Atlas (85.6% vs. 85.3%) and Draco (83.3% vs. 78.5%). With tools, HLE rises from 46.5% → 59.4% (+12.9 points).
📝 Benchmark descriptions
- GDPval-AA v2 — Artificial Analysis, Sep 19, 2026.
- τ³-Banking — Multi-turn banking agent benchmark.
- AutomationBench-AA / Public — Agentic automation tasks, private and public splits.
- AA-Briefcase — Elo score; agentic knowledge work across data science, product, banking, heavy industry.
- Toolathlon-Verified — 108 expert-authored tasks; multi-app workflows averaging ~20 turns.
- MCP-Atlas — 1,000 tasks across 36 real MCP servers and 220 tools; 3–6 tool calls per task.
- PresentBench — 238 slide-generation instances; avg 54.1 rubric checklist items per instance.
- JobBench — 130 tasks across 35 occupations; avg 35.6 binary rubric criteria per task.
- Apex-Agents — Long-horizon tasks in investment banking, consulting, and corporate law.
- DRACO — Cross-domain deep research benchmark; LLM-as-judge with binary rubric verdicts.
- BrowseComp — 1,266 hard-to-find web information retrieval questions.
- HLE w/ tools — Humanity's Last Exam with tool augmentation.
💰 Finance & Professional Work
| Benchmark | Step-5-Preview | GPT-6 Astra | Fable 5.1 | Claude Opus 5 | Kimi K3 | GLM-5.3 |
|---|---|---|---|---|---|---|
| FinStepBench-LiveSearch | 74.5% | 74.5% | — | 76.2% 🥇 | 70.9% | 73.3% |
| FinStepBench-CorporateValuation | 60.6% | 77.3% 🥇 | — | 69.7% | 60.6% | 56.1% |
| FinStepBench-FinanceDR | 55.8% | 45.0% | — | 59.1% 🥇 | 48.9% | 53.3% |
| FrontierFinance | 66.4% | 55.0% | — | 69.7% 🥇 | 62.6% | 64.1% |
| OfficeQA Pro | 60.3% | 67.7% 🥇 | — | 64.7% | 62.6% | 59.1% |
| Spreadsheet v2 | 29.4% | 31.4% | — | 32.8% 🥇 | 31.9% | 30.5% |
| GDP.pdf | 14.8% | 31.0% 🥇 | 26.2% | 21.6% | 22.0% | 11.2% |
On FrontierFinance Step-5-Preview scores 66.4%, ahead of GPT-6 Astra by +11.4, Kimi K3 by +3.8, and GLM-5.3 by +2.3. On FinStepBench-FinanceDR it reaches 55.8%, ahead of GPT-6 Astra (+10.8), Kimi K3 (+6.9), and GLM-5.3 (+2.5).
📝 Benchmark descriptions
- FinStepBench-LiveSearch — StepFun internal; live web search for financial research.
- FinStepBench-CorporateValuation — StepFun internal; corporate valuation tasks.
- FinStepBench-FinanceDR — StepFun internal; deep-research finance workflows with source traceability, reproducible assumptions, auditable reports.
- FrontierFinance — Finance reasoning benchmark.
- OfficeQA Pro — Databricks; 90 questions over large enterprise financial document collections.
- Spreadsheet v2 — SpreadsheetBench 2; end-to-end business spreadsheet workflows. Best model ≈34.89%.
- GDP.pdf — Document-heavy visual QA over PDFs.
👁️ Multimodal
| Benchmark | Step-5-Preview | GPT-6 Astra | Fable 5.1 | Claude Opus 5 | Kimi K3 | GLM-5.3 |
|---|---|---|---|---|---|---|
| MMMU-Pro | 76.0% | 87.0% 🥇 | — | 85.0% | 81.0% | — |
| GDP.pdf | 14.8% | 31.0% 🥇 | 26.2% | 21.6% | 22.0% | 11.2% |
Multimodal reasoning is the primary improvement target for the next release. MMMU-Pro at 76.0% trails GPT-6 Astra (87.0%), Claude Opus 5 (85.0%), and Kimi K3 (81.0%).
📝 Benchmark descriptions
- MMMU-Pro — Robust multimodal benchmark; filtered text-only questions, augmented candidates, vision-only settings. Significantly harder than MMMU.
- GDP.pdf — Document-heavy visual QA over PDFs.
🏅 Category Standing
| Domain | Step-5-Preview Standing | Highlight |
|---|---|---|
| Long-Context Reasoning | 🥈 Near-tied for #1 | AA-LCR v1.1 88.3% |
| Coding / SWE | 🥇 #1 open-weight | DeepSWE v1.1 67.7%, StepCodeBench 49.0%, ProgramBench 80.5% |
| Security / Cyber | 🥇 #1 in set | CyberGym 84.7% |
| Finance | 🥈 #2 overall, #1 open-weight | FrontierFinance 66.4%, FinanceDR 55.8% |
| Terminal Agents | 🥈 #2 open-weight | Terminal-Bench 4 33.3% |
| Tool Use | 🥉 Competitive | MCP-Atlas 85.6%, HLE w/ tools 59.4% |
| General Knowledge | Top tier | GPQA Diamond 93.5%, HLE 46.5% |
| Multimodal | Trailing | MMMU-Pro 76.0% |
📝 Benchmark methodology notes
- DeepSWE v1.1 was evaluated using the SWE-agent harness with
temperature=1.0andtop_p=0.95. - StepCodeBench achieved 49.0% avg@4.
- GDPval-AA v2 results are from Artificial Analysis as of September 19, 2026.
- AA-LCR v1.1: statistically tied with Kimi K3 (88.7%).
- Terminal-Bench 4: 33.3% vs. Kimi K3 ~12.6% (2.6×) and DeepSeek V4.1 Flash 26.8% (1.24×).
- SciCode: 58.9%, above GPT-6 Astra (56.5%) and Claude Opus 5 (56.4%).
- Multimodal: MMMU-Pro and GDP.pdf run with the unified multimodal encoder at default resolution.
- Output Speed: 99.8 tokens/sec (GLM-5.3: 72.1 tokens/sec).
- Time to First Token: 2.96 seconds (GLM-5.3: 2.99s; Claude Opus 5: 56.84s at max effort).
- Missing entries (—) reflect benchmarks that were not publicly reported for that model at the time of writing.
🤖 Agentic Capabilities
24-Hour Autonomous GPU Kernel Optimization
In a landmark demonstration of sustained agentic execution, Step-5-Preview was tasked with autonomously optimizing an H100 GPU kernel for up to 24 consecutive hours. The model:
- Independently modified code
- Ran tests and compared results
- Iterated based on performance outcomes
- Reached 508 TFLOPS after approximately 22 hours
For comparison, Claude Opus 5 achieved 493 TFLOPS in the same experiment. This demonstrates Step-5-Preview's ability to sustain productive work over extended periods without human intervention.
Automated Post-Training Experiments
In another 24-hour experiment, Step-5-Preview autonomously improved the accuracy of Qwen3-30B-A3B on AIME24 from 53.3% to 60% through automated post-training experiments. This showcases the model's capacity for self-directed research and optimization.
Long-Horizon Agent Workflows
The model is specifically optimized for agent workflows that require:
- Searching and information retrieval
- Running code and processing tool returns
- Multi-turn tool calls with sustained execution
- Iterative refinement based on intermediate results
- Self-correction and error recovery over thousands of steps
💼 Real-World Use Cases
StepFun demonstrated the model's capabilities across several complex, real-world projects:
- ESP32 Development Board Modifications: Executed development tasks for over 3 hours, demonstrating hardware programming capabilities.
- Front-End Design with 3D Asset Generation: Full-stack development workflows including visual design.
- Full-Process Financial Research: End-to-end investment research workflows, from data gathering to report generation.
- Software Engineering: Comprehensive coding tasks beyond traditional code generation, including front-end, visual development, and programmable hardware scenarios.
- Autonomous Research Assistant: Capable of reading papers, running experiments, and summarizing findings.
- Customer Support Automation: Handles multi-turn conversations with tool calls to internal systems.
⚡ Quickstart
Installation
pip install transformers>=4.56.0
pip install torch>=2.4.0
pip install accelerate
For video/image support:
pip install av pillow
Basic Usage with Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
model_id = "SHSLab/Step-5-Preview-BF16"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
model_id,
trust_remote_code=True,
device_map="auto",
torch_dtype="bfloat16",
)
messages = [
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "Explain the significance of the Pareto Frontier in AI scaling."},
]
inputs = tokenizer.apply_chat_template(
messages,
add_generation_prompt=True,
return_tensors="pt",
).to(model.device)
outputs = model.generate(
inputs,
max_new_tokens=1024,
temperature=0.7,
top_p=0.95,
reasoning_effort="high", # low / medium / high / xhigh
)
response = tokenizer.decode(outputs[0][inputs.shape[1]:], skip_special_tokens=True)
print(response)
Multimodal (Image + Video) Usage
from transformers import AutoProcessor
processor = AutoProcessor.from_pretrained(model_id, trust_remote_code=True)
messages = [
{
"role": "user",
"content": [
{"type": "image", "url": "https://example.com/image.jpg"},
{"type": "video", "url": "https://example.com/video.mp4"},
{"type": "text", "text": "Describe the scene and summarize the video."},
],
}
]
inputs = processor.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt")
# ... generate as above
Tool Calling
tools = [
{
"type": "function",
"function": {
"name": "get_weather",
"parameters": {
"type": "object",
"properties": {"city": {"type": "string"}},
"required": ["city"],
},
},
}
]
messages = [{"role": "user", "content": "What's the weather in Tokyo?"}]
inputs = tokenizer.apply_chat_template(
messages,
tools=tools,
add_generation_prompt=True,
return_tensors="pt",
).to(model.device)
outputs = model.generate(inputs, max_new_tokens=256, reasoning_effort="medium")
print(tokenizer.decode(outputs[0][inputs.shape[1]:], skip_special_tokens=True))
🚢 Deployment
vLLM
vllm serve SHSLab/Step-5-Preview-BF16 \
--trust-remote-code \
--tensor-parallel-size 8 \
--max-model-len 1000000 \
--enable-reasoning \
--reasoning-parser stepfun
SGLang
python -m sglang.launch_server \
--model-path SHSLab/Step-5-Preview-BF16 \
--trust-remote-code \
--tp 8 \
--context-length 1000000 \
--reasoning-parser stepfun
OpenAI-Compatible API
from openai import OpenAI
client = OpenAI(
api_key="YOUR_STEP_API_KEY",
base_url="https://api.stepfun.com/v1",
)
response = client.chat.completions.create(
model="step-5-preview",
messages=[{"role": "user", "content": "Write a Python function to merge two sorted lists."}],
reasoning_effort="high",
max_tokens=2048,
)
print(response.choices[0].message.content)
📦 Recommended Deployment Configurations
• BF16: 8× H100 80GB (tensor parallel) • FP8: 4× H100 80GB (coming soon) • Context length: Up to 1M tokens • Reasoning parser: Use
stepfunfor vLLM / SGLang
📈 Evaluation
Step-5-Preview was evaluated on a comprehensive suite of public and internal benchmarks. All evaluations used the model's high reasoning effort setting unless otherwise noted.
| Benchmark | Score | Notes |
|---|---|---|
| DeepSWE v1.1 | 67.7% | SWE-agent harness, temp=1.0, top_p=0.95 |
| StepCodeBench | 49.0% | avg@4 |
| ProgramBench | 80.5% | Rebuild programs from binary + docs; 200 tasks, 248K+ behavioral tests |
| Terminal-Bench 2.1 | 85.0% | Verified refresh of TB 2.0; 89 curated terminal tasks |
| Terminal-Bench 4 | 33.3% | 2.6× Kimi K3 |
| CyberGym | 84.7% | Best in comparison set |
| SciCode | 58.9% | Above GPT-6 Astra (56.5%) and Opus 5 (56.4%) |
| RoadmapBench | 54.3% | 115 long-horizon coding tasks; median ~3,700 LOC changed |
| SWE-Marathon v1.1 | 72.7% | Ultra-long-horizon SWE; 20 realistic multi-hour tasks |
| MLS-Bench-Lite | 40.5% | 30-task subset; inventing generalizable ML methods |
| SWE-Atlas-QnA | 63.6% | 124 deep code-comprehension tasks |
| SWE-Atlas-Test-writing | 50.8% | 90 test-writing tasks across production repos |
| StepCode-Bench-Daily | 64.9% | StepFun internal; 553 repos, 9 task types, 33 languages |
| StepCode-Bench-General | 65.0% | StepFun internal; general-difficulty slice |
| Agents' Last Exam (ALE-CLI) | 29.5% | Linux-only CLI subset; 40 industry subfields |
| GDPval-AA v2 | 1571 | Artificial Analysis, Sep 19, 2026 |
| AA-Briefcase | 1417 | Elo; agentic knowledge work across industries |
| Toolathlon-Verified | 74.1% | 108 expert-authored tasks; ~20 turns average |
| MCP-Atlas | 85.6% | 1,000 tasks across 36 MCP servers, 220 tools |
| PresentBench | 76.8% | 238 slide-generation instances; avg 54.1 rubric items |
| JobBench | 59.0% | 130 tasks across 35 occupations |
| Apex-Agents | 37.8% | Long-horizon IB, consulting, corporate law tasks |
| DRACO | 83.3% | Cross-domain deep research benchmark |
| BrowseComp | 88.7% | 1,266 hard web information retrieval questions |
| HLE w/ tools | 59.4% | +12.9 pts over no-tools HLE |
| FrontierFinance | 66.4% | +11.4 pts over GPT-6 Astra |
| FinStepBench-LiveSearch | 74.5% | Tied with GPT-6 Astra |
| FinStepBench-CorporateValuation | 60.6% | Tied with Kimi K3 |
| FinStepBench-FinanceDR | 55.8% | Deep-research finance workflows |
| OfficeQA Pro | 60.3% | 90 questions over enterprise financial documents |
| Spreadsheet v2 | 29.4% | SpreadsheetBench 2; best model ≈34.89% |
| GDP.pdf | 14.8% | Known weak spot |
| GPQA Diamond | 93.5% | 198 PhD-level science MCQs; human expert avg 81% |
| HLE | 46.5% | 59.4% with tools |
| AA-LCR v1.1 | 88.3% | Tied with Kimi K3 (88.7%) |
| CritPt | 20.9% | 71 unpublished research-level physics challenges |
| MMMU-Pro | 76.0% | Robust multimodal benchmark; harder than MMMU |
| Output Speed | 99.8 tokens/sec | GLM-5.3: 72.1 tokens/sec |
| Time to First Token | 2.96s | GLM-5.3: 2.99s; Claude Opus 5: 56.84s (max effort) |
⚠️ Limitations
- Knowledge Cutoff: The model's knowledge is current up to mid-2026. It may not be aware of events after that date.
- Hallucination: Like all large language models, Step-5-Preview can generate plausible but incorrect information, especially in domains with sparse training data.
- Long Context Degradation: While the model supports 1M tokens, performance may degrade for extremely long contexts beyond 500K tokens in certain tasks.
- Tool Use Reliability: Tool calling is highly capable but not infallible. Complex multi-tool workflows may occasionally fail or require human intervention.
- Multimodal Limitations: Video understanding is limited to clips under 5 minutes and 128 MB. Extremely high-resolution images may be downscaled.
- Language Coverage: While multilingual, the model is primarily optimized for English and Chinese. Performance in other languages may vary.
⚖️ Ethical Considerations
StepFun is committed to the responsible development and deployment of AI. We have taken the following measures:
- Safety Alignment: The model was fine-tuned with RLHF to refuse harmful requests and promote helpful, honest, and harmless behavior.
- Bias Mitigation: Training data was filtered to reduce harmful stereotypes and biases. However, residual biases may exist.
- Transparency: We provide detailed model cards and benchmark results to enable informed use.
- License Restrictions: The StepFun Community License prohibits certain high-risk uses, including autonomous weapons, surveillance, and malicious cyber activities.
- Content Provenance: We encourage users to clearly label AI-generated content and to use the model ethically.
We urge all users to consider the ethical implications of their applications and to implement appropriate safeguards.
🖥️ Hardware Requirements
| Precision | Minimum GPU Memory | Recommended GPU Configuration |
|---|---|---|
| BF16 | 1.2 TB | 8× H100 80GB (tensor parallel) |
| FP8 | 600 GB | 4× H100 80GB (tensor parallel) |
| INT4 | 300 GB | 4× A100 80GB (tensor parallel) |
For inference with 1M context, additional memory is required for KV cache. We recommend using paged attention and offloading techniques available in vLLM and SGLang.
⚡ Performance Metrics
| Metric | Value |
|---|---|
| Output Speed | 99.8 tokens/sec |
| Time to First Token (TTFT) | 2.96 seconds |
| Context Window | 1,000,000 tokens |
| Max Output Tokens | 32,768 (default), configurable up to 131,072 |
| Reasoning Effort Modes | low, medium, high, xhigh |
| Tool Calling Latency | < 500 ms for simple calls |
Measured on 8× H100 80GB with vLLM, batch size 1, BF16.
📚 Citation
If you use Step-5-Preview in your research, please cite:
@misc{stepfun2026step5preview,
title = {Step-5-Preview: A 600B Sparse MoE Foundation Model for Real-World Agentic Work},
author = {StepFun Team},
year = {2026},
howpublished = {\url{https://huggingface.co/SHSLab/Step-5-Preview-BF16}},
note = {Released September 20, 2026}
}
📜 License
Step-5-Preview is released under the StepFun Community License. See the LICENSE file for full terms.
⚠️ Usage Restrictions
• Commercial use is permitted under the StepFun Community License. • Redistribution must include the license and attribution. • See LICENSE for full details.
📬 Contact
- Hugging Face: SHSLab
- GitHub: github.com/stepfun-ai
- Discord: Join our Discord
- Email: opensource@stepfun.com
- Website: stepfun.com
Built with ❤️ by StepFun
- Downloads last month
- -