Check my GPU

LLMs ·

Step 3.5 Flash (196B-A11B)

Parameters
199B (11.0B active)
Architecture
Mixture of experts
Max context
256K tokens
License
apache-2.0
Released
2026-02

How to run it

Quantized files · GGUF

Works in LM Studio, llama.cpp, Jan and Ollama. No sign-in needed.

huggingface.co/bartowski/stepfun-ai_Step-3.5-Flash-GGUF (opens in a new tab)

ollama run hf.co/bartowski/stepfun-ai_Step-3.5-Flash-GGUF

Original weights · Hugging Face

Full-precision files from StepFun, for fine-tuning, vLLM, Transformers.

huggingface.co/stepfun-ai/Step-3.5-Flash (opens in a new tab)

No sign-in needed.

Official repo · GitHub

Reference code, inference scripts and docs from StepFun. The weights themselves download from Hugging Face.

github.com/stepfun-ai/Step-3.5-Flash (opens in a new tab)

git clone https://github.com/stepfun-ai/Step-3.5-Flash

Links checked 5 days ago against Hugging Face, Ollama and GitHub.

Memory needed

Weights plus KV cache and runtime overhead. Leave a little extra free for the operating system and other apps.

KV cache: sliding-window hybrid: the KV estimate counts the global-attention layers plus a small allowance for the sliding-window layers; KV heads assumed to be 8 (grouped attention).

Quantization Quality Weights Total at 8K context Total at 32K context Download
Q2_K Low 79.8 GB 80.8 GB 82.1 GB GGUF · folder (opens in a new tab)
Q3_K_M Low 97.7 GB 98.7 GB 100 GB GGUF · folder (opens in a new tab)
Q4_K_M Good 122 GB 123 GB 124 GB GGUF · folder (opens in a new tab)
Q5_K_M High 142 GB 143 GB 144 GB GGUF · folder (opens in a new tab)
Q6_K High 163 GB 165 GB 166 GB GGUF · folder (opens in a new tab)
Q8_0 Very high 211 GB 212 GB 214 GB GGUF · folder (opens in a new tab)
F16 Full 399 GB 400 GB 401 GB Original weights (opens in a new tab)

Best hardware for Step 3.5 Flash (196B-A11B)

The least expensive GPU in our catalog that runs it without offloading is the NVIDIA RTX PRO 6000 Blackwell (launch price $8,565) at Q2_K.

Grade · points Hardware Quant Memory Speed Fit
94 ptsExcellent (S) · 94 of 100 points: Model quality 39.3/45, Memory fit 25.0/25, Speed 14.9/15, Memory use 15.0/15 Apple M3 Ultra (256 GB) Q4_K_M 138 GB / 192 GB 39.4 tok/s Runs well
93 ptsExcellent (S) · 93 of 100 points: Model quality 39.3/45, Memory fit 24.0/25, Speed 14.7/15, Memory use 15.0/15 Apple M2 Ultra (192 GB) Q4_K_M 134 GB / 144 GB 38.4 tok/s Tight fit
88 ptsExcellent (S) · 88 of 100 points: Model quality 39.3/45, Memory fit 25.0/25, Speed 14.9/15, Memory use 9.2/15 Apple M3 Ultra (512 GB) Q4_K_M 153 GB / 384 GB 39.4 tok/s Runs well
84 ptsGreat (A) · 84 of 100 points: Model quality 34.8/45, Memory fit 24.0/25, Speed 10.5/15, Memory use 15.0/15 NVIDIA GB10 (128 GB) Q3_K_M 108 GB / 115 GB 19.7 tok/s Tight fit
82 ptsGreat (A) · 82 of 100 points: Model quality 29.0/45, Memory fit 24.0/25, Speed 15.0/15, Memory use 15.0/15 Apple M4 Max (128 GB) Q2_K 88.5 GB / 96.0 GB 43.1 tok/s Tight fit
82 ptsGreat (A) · 82 of 100 points: Model quality 29.0/45, Memory fit 24.0/25, Speed 15.0/15, Memory use 15.0/15 NVIDIA RTX PRO 6000 Blackwell Q2_K 85.6 GB / 96.0 GB 164 tok/s Tight fit
78 ptsGreat (A) · 78 of 100 points: Model quality 29.0/45, Memory fit 24.0/25, Speed 10.7/15, Memory use 15.0/15 AMD Ryzen AI Max+ 395 (128 GB) Q2_K 88.5 GB / 96.0 GB 20.2 tok/s Tight fit

Check it on your own hardware

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.