Step 3.5 Flash
- In Input modalities: Text Out Output modalities: Text
- Reasoning· low, high
- Tool use
- Open weights
Prices
2 offerings · prices as listedstep-3.5-flash| Conditions/ 1M tokens | Input | Cache read | Output |
|---|---|---|---|
| Standard | $0.10 | $0.020 | $0.30 |
Prices from StepFun: pricing details (opens in a new tab) (checked Nov 2, 2026)
step-3.5-flash| Conditions/ 1M tokens | Input | Cache read | Output |
|---|---|---|---|
| Standard | ¥0.70 | ¥0.14 | ¥2.10 |
Prices from 阶跃星辰：定价明细 (opens in a new tab) (checked Nov 2, 2026)
Standard: the price a request with no conditions pays (the headline price elsewhere on the site).
Details
- Knowledge cutoff
- Jan 2025
- Licence
- Apache 2.0
- Family
- step-flash
- Model ID
Not stated on StepFun’s pages: max input, reasoning can be off, max thinking budget, structured output, built-in tools, aliases.
Run locally
Weights: stepfun-ai/Step-3.5-Flash- Parameters 199.4B
- Active 11B · 288 experts, 8 per token
- Native precision BF16
- Layers 45
- KV cache per token 180 KB
- Config positions 256K (config.json max_position_embeddings)
- API context 250K (maker’s API)
- Needs
- ≈ 105.0 GB (weights 98.6 + KV cache 1.4 + overhead 4.9)
- Usable
- 21.6 GB (90% of its memory)
- Try
- IQ3_XXS runs (needs system memory), but below 4 bits quality drops noticeably
| Precision | Weights | Needs at 8K | On your hardware | Longest context |
|---|---|---|---|---|
| BF16 (native)maker’s file | 371.4 GB | ≈ 391.4 | ✕Doesn’t fit | — |
| Q8_0 | 197.3 GB | ≈ 208.6 | ✕Doesn’t fit | — |
| FP8maker’s file | 194.2 GB | ≈ 205.4 | ✕Doesn’t fit | — |
| Q6_K | 152.5 GB | ≈ 161.5 | ✕Doesn’t fit | — |
| Q5_K_M | 132.1 GB | ≈ 140.1 | ✕Doesn’t fit | — |
| Q5_0 | 127.7 GB | ≈ 135.5 | ✕Doesn’t fit | — |
| Q4_K_M | 112.8 GB | ≈ 119.9 | ✕Doesn’t fit | — |
| Q4_0 | 104.5 GB | ≈ 111.1 | ✕Doesn’t fit | — |
| IQ4_XS | 98.6 GB | ≈ 105.0 | ✕Doesn’t fit | — |
| Q3_K_Mquality drops | 90.8 GB | ≈ 96.7 | ✕Doesn’t fit | — |
| IQ3_XXSquality drops | 71.0 GB | ≈ 76.0 | ○Needs system memory 13–20 tok/s | — |
| Q2_Kquality drops | 69.6 GB | ≈ 74.5 | ○Needs system memory 14–20 tok/s | — |
| IQ2_XXSquality drops | 47.8 GB | ≈ 51.6 | ○Needs system memory 30–44 tok/s | — |
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): GPU NVIDIA H200 SXM; unified memory Apple M5 Ultra (64-core GPU) · 256 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 2026-11-06. How it’s estimated
Cited scores
From third-party leaderboards, not measured by us- LMArena Text 1,393
#105 of 163 listed models · CI 1,390–1,397 · 60,305 votes
- LMArena Text · Chinese 1,432
#97 of 161 listed models · CI 1,421–1,442 · 3,667 votes
- LMArena Text · Coding 1,449
#96 of 163 listed models · CI 1,444–1,455 · 15,184 votes
- LMArena Text · Creative writing 1,345
#109 of 163 listed models · CI 1,338–1,352 · 9,760 votes
- LMArena (Arena), leaderboard-dataset on Hugging Face, licensed CC BY 4.0 · License · as of Oct 28, 2026
- changes we made
Changes
All changes →No changes recorded yet.
Sources
- StepFun: pricing details (opens in a new tab) platform.stepfun.ai · checked Nov 2, 2026
- 阶跃星辰：定价明细in Chinese (opens in a new tab) platform.stepfun.com · checked Nov 2, 2026
- Step 3.5 Flash (opens in a new tab) platform.stepfun.ai · checked Nov 1, 2026
- StepFun: reasoning models (opens in a new tab) platform.stepfun.ai · checked Nov 2, 2026
- stepfun-ai/Step-3.5-Flash model card (opens in a new tab) huggingface.co · checked Nov 1, 2026
- models.dev: stepfun/step-3.5-flash (community data, MIT) (opens in a new tab) github.com · checked Oct 19, 2026
- stepfun-ai/Step-3.5-Flash model card (opens in a new tab) huggingface.co · checked Nov 1, 2026
- stepfun-ai/Step-3.5-Flash model card (opens in a new tab) huggingface.co · checked Oct 30, 2026
- stepfun-ai/Step-3.5-Flash files (opens in a new tab) huggingface.co · checked Nov 3, 2026
- stepfun-ai/Step-3.5-Flash-FP8 files (opens in a new tab) huggingface.co · checked Nov 4, 2026
- stepfun-ai/Step-3.5-Flash config.json (opens in a new tab) huggingface.co · checked Nov 1, 2026
- stepfun-ai/Step-3.5-Flash README.md (opens in a new tab) huggingface.co · checked Nov 2, 2026
- Step 3.5 Flash (opens in a new tab) platform.stepfun.ai · checked Nov 1, 2026
Some values on this page are still waiting to be checked against official pages: a filled dot is verified, a hollow one is not yet (with or without a source on file). How we verify data
StepFun · China · stepfun.com
Related models
Snapshots, the same family and other models from this provider