Step 5 Preview
Step 5 Preview is a StepFun model designed for agents handling software development, document analysis, and research. It can work through tasks in stages using tools connected by the integrating application. Summary written by AI Index from StepFun’s own pages.
- In Input modalities: TextImageVideo Out Output modalities: Text
- Reasoning· low, medium, high
- Tool use
- Structured output
- Open weights
Prices
2 offerings · prices as listed| Conditions/ 1M tokens | Input | Cache read | Output |
|---|---|---|---|
| Standard | $1.00 | $0.10 | $2.70 |
Prices from StepFun: pricing details (opens in a new tab) (checked Nov 7, 2026)
| Conditions/ 1M tokens | Input | Cache read | Output |
|---|---|---|---|
| Standard | ¥7.00 | ¥0.70 | ¥20.00 |
Prices from 阶跃星辰:定价明细 (opens in a new tab) (checked Nov 9, 2026)
Standard: the price a request with no conditions pays (the headline price elsewhere on the site).
Details
- Licence
- StepFun Model License 1.0
- Family
- step
- Model ID
Not stated on StepFun’s pages: reasoning can be off, max thinking budget, built-in tools, knowledge cutoff, aliases.
Run locally
Weights: stepfun-ai/Step-5-Preview- Parameters 604.3B
- Dense
- Native precision BF16
- Layers 92
- KV cache per token 276 KB
- API context 977K (maker’s API)
- Needs
- ≈ 316.1 GB (weights 299.0 + KV cache 2.2 + overhead 15.0)
- Usable
- 21.6 GB (90% of its memory)
- Try
- it’s too large for this hardware; see the smallest that runs it below
| Precision | Weights | Needs at 8K | On your hardware | Longest context |
|---|---|---|---|---|
| BF16 (native)maker’s file | 605.6 GB | ≈ 638.1 | ✕Doesn’t fit | — |
| Q8_0 | 598.0 GB | ≈ 630.1 | ✕Doesn’t fit | — |
| FP8 | 562.8 GB | ≈ 593.1 | ✕Doesn’t fit | — |
| Q6_K | 462.2 GB | ≈ 487.5 | ✕Doesn’t fit | — |
| Q5_K_M | 400.3 GB | ≈ 422.5 | ✕Doesn’t fit | — |
| Q5_0 | 386.9 GB | ≈ 408.5 | ✕Doesn’t fit | — |
| Q4_K_M | 341.9 GB | ≈ 361.2 | ✕Doesn’t fit | — |
| NVFP4maker’s file | 367.1 GB | ≈ 387.6 | ✕Doesn’t fit | — |
| Q4_0 | 316.6 GB | ≈ 334.6 | ✕Doesn’t fit | — |
| IQ4_XS | 299.0 GB | ≈ 316.1 | ✕Doesn’t fit | — |
| Q3_K_Mquality drops | 275.1 GB | ≈ 291.0 | ✕Doesn’t fit | — |
| IQ3_XXSquality drops | 215.3 GB | ≈ 228.2 | ✕Doesn’t fit | — |
| Q2_Kquality drops | 211.1 GB | ≈ 223.8 | ✕Doesn’t fit | — |
| IQ2_XXSquality drops | 144.9 GB | ≈ 154.3 | ✕Doesn’t fit | — |
Click a row to choose a precision. “Maker’s file” sizes are the maker’s own files; others are parameters × bits per weight.
Compared with other hardware
Each at the highest precision it runs. Click one for details.
Smallest that runs it (Q4_K_M, 8K): no single listed GPU; unified memory Apple M5 Ultra (80-core GPU) · 512 GB.
Run commands
The maker publishes no files at this precision. Choose its native precision or one of its own builds, or quantise it yourself with llama.cpp.
Memory to fine-tune
Batch 1, 2,048-token sequences, gradient checkpointing; LoRA and QLoRA train rank-16 adapters on the attention projections.
Estimates; facts from the maker’s Hugging Face repositories, as of 2026-11-21. How it’s estimated
Cited scores
From third-party leaderboards, not measured by us- LMArena Text 1,448
#53 of 163 listed models · CI 1,442–1,455 · 12,369 votes
- LMArena Text · Chinese 1,499
#44 of 161 listed models · CI 1,480–1,518 · 989 votes
- LMArena Text · Coding 1,505
#46 of 163 listed models · CI 1,493–1,516 · 2,954 votes
- LMArena Text · Creative writing 1,412
#57 of 163 listed models · CI 1,399–1,424 · 2,649 votes
- LMArena (Arena), leaderboard-dataset on Hugging Face, licensed CC BY 4.0 · License · as of Nov 8, 2026
- changes we made
News
All news →- Oct 12, 2026Pricing StepFun raises Step 5 Preview cache read prices by 100% in China Step 5 Preview on StepFun API (CNY): cache read ¥0.35 → ¥0.70 per 1M tokens.
- Oct 10, 2026Pricing StepFun raises Step 5 Preview cache read prices by 100% Step 5 Preview on StepFun API (USD): cache read $0.05 → $0.10 per 1M tokens.
Changes
All changes →- Cache read price raised 100% ¥0.35 → ¥0.70 / 1M tokens · StepFun API, China
- Cache read price raised 100% $0.050 → $0.10 / 1M tokens · StepFun API
Sources
- StepFun: pricing details (opens in a new tab) platform.stepfun.ai · checked Nov 7, 2026
- 阶跃星辰:定价明细in Chinese (opens in a new tab) platform.stepfun.com · checked Nov 9, 2026
- Step 5 Preview (opens in a new tab) platform.stepfun.ai · checked Nov 8, 2026
- Step 5 Preview (opens in a new tab) platform.stepfun.ai · checked Nov 20, 2026
- stepfun-ai/Step-5-Preview model card (opens in a new tab) huggingface.co · checked Nov 6, 2026
- stepfun-ai/Step-5-Preview files (opens in a new tab) huggingface.co · checked Nov 7, 2026
- stepfun-ai/Step-5-Preview-NVFP4 files (opens in a new tab) huggingface.co · checked Nov 8, 2026
- stepfun-ai/Step-5-Preview config.json (opens in a new tab) huggingface.co · checked Nov 4, 2026
- Step 5 Preview (opens in a new tab) huggingface.co · checked Nov 5, 2026
StepFun · China · stepfun.com
Related models
Snapshots, the same family and other models from this provider
- Step 3.7 Flash
- Step Audio 2
- Step 1o Audio
- Step 1o Turbo Vision
- Step 3.5 Flash
- Step 3.5 Flash 2603
- Step Audio R1.5
- StepAudio 2.5 Chat
- GPT-6.2 Astra vs Step 5 Preview
- Qwen4 Max vs Step 5 Preview
- LongCat-2.5 vs Step 5 Preview
- Hy4 Preview vs Step 5 Preview
- MiMo-V2.6 Pro vs Step 5 Preview
- Mistral Large 4 vs Step 5 Preview
- Muse Wonder vs Step 5 Preview
- Seed 2.1 Pro vs Step 5 Preview