Step 3.5 Flash needs about 134.8 GB of memory at Q4_K_M (a 119.9 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 28 quants, from 0.5 GB to 209.4 GB (Q8_0).
Step 3.5 Flash – a 196B sparse MoE model with 256K context
Step 3.5 Flash is a 196B parameter sparse Mixture of Experts model engineered for deep reasoning and agentic tasks. It supports a 256K token context window and is optimized for local deployment on consumer hardware. The model is released under Apache‑2.0.
Good at
- Deep reasoning at high speed
- Agentic task execution
- Long‑context handling
Limits
- Requires high‑end GPU for full performance
- No vision support
- Large memory footprint for full quant
Best for
- Complex coding assistants
- Long‑form reasoning workflows
- Agentic automation scripts
Who it’s for. Developers and researchers needing a powerful open‑source foundation model for reasoning and agentic applications. The model is suitable for local deployment on high‑end consumer GPUs and CPUs.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.5 GB | 1.0 GB | Fits | Get file → | |
| IQ1_S | 41.3 GB | 46.8 GB | Too big | Get file → | |
| IQ1_M | 43.1 GB | 48.8 GB | Too big | Get file → | |
| IQ2_XXS | 48.0 GB | 54.3 GB | Too big | Get file → | |
| IQ2_XS | 55.2 GB (2 parts) | 62.3 GB | Too big | Get file → | |
| IQ2_S | 55.5 GB (2 parts) | 62.7 GB | Too big | Get file → | |
| IQ2_M | 62.9 GB (2 parts) | 71.0 GB | Too big | Get file → | |
| Q2_K | 69.6 GB (2 parts) | 78.4 GB | Too big | Get file → | |
| Q2_K_L | 70.1 GB (2 parts) | 79.0 GB | Too big | Get file → | |
| IQ3_XXS | 78.2 GB (2 parts) | 88.1 GB | Too big | Get file → | |
| IQ3_XS | 81.0 GB (3 parts) | 91.3 GB | Too big | Get file → | |
| Q3_K_S | 86.0 GB (3 parts) | 96.8 GB | Too big | Get file → | |
| IQ3_M | 90.2 GB (3 parts) | 101.5 GB | Too big | Get file → | |
| Q3_K_M | 90.2 GB (3 parts) | 101.6 GB | Too big | Get file → | |
| Q3_K_L | 93.7 GB (3 parts) | 105.4 GB | Too big | Get file → | |
| Q3_K_XL | 94.2 GB (3 parts) | 106.0 GB | Too big | Get file → | |
| IQ4_XS | 105.6 GB (3 parts) | 118.8 GB | Too big | Get file → | |
| IQ4_NL | 111.6 GB (3 parts) | 125.5 GB | Too big | Get file → | |
| Q4_0 | 113.4 GB (3 parts) | 127.5 GB | Too big | Get file → | |
| Q4_K_S | 115.6 GB (3 parts) | 130.0 GB | Too big | Get file → | |
| Q4_K_M | 119.9 GB (4 parts) | 134.8 GB | Too big | Get file → | |
| Q4_K_L | 120.3 GB (4 parts) | 135.3 GB | Too big | Get file → | |
| Q4_1 | 123.6 GB (4 parts) | 139.0 GB | Too big | Get file → | |
| Q5_K_S | 135.9 GB (4 parts) | 152.7 GB | Too big | Get file → | |
| Q5_K_M | 140.2 GB (4 parts) | 157.5 GB | Too big | Get file → | |
| Q5_K_L | 140.5 GB (4 parts) | 157.9 GB | Too big | Get file → | |
| Q6_K | 161.9 GB (5 parts) | 181.9 GB | Too big | Get file → | |
| Q8_0 | 209.4 GB (6 parts) | 235.0 GB | Too big | Get file → |
For 8 GB VRAM, use IQ1_S (41.3 GB file, 46.8 GB memory at 8K context). For 16 GB VRAM, IQ2_XS (55.19 GB file, 62.3 GB memory) or IQ2_S (55.51 GB file, 62.7 GB memory). For 24 GB VRAM, Q3_K_M (90.23 GB file, 101.6 GB memory) or Q4_K_M (119.94 GB file, 134.8 GB memory).
What GPU do you need for Step 3.5 Flash?
These figures add the file size, an fp16 KV cache at an 8K context window and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: the best quant that fits is the main file (0.5 GB file, about 1.0 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (0.5 GB file, about 1.0 GB in use).
- 8 GB VRAM: the best quant that fits is the main file (0.5 GB file, about 1.0 GB in use).
- 12 GB VRAM: the best quant that fits is the main file (0.5 GB file, about 1.0 GB in use).
- 16 GB VRAM: the best quant that fits is the main file (0.5 GB file, about 1.0 GB in use).
- 24 GB VRAM: the best quant that fits is the main file (0.5 GB file, about 1.0 GB in use).
- 32 GB VRAM: the best quant that fits is the main file (0.5 GB file, about 1.0 GB in use).
- 48 GB VRAM: the best quant that fits is IQ1_S (41.3 GB file, about 46.8 GB in use).
Which Step 3.5 Flash quant should you choose?
Q4_K_M is the usual starting point for Step 3.5 Flash: it balances file size and quality well. Here it is a 119.9 GB file.
The Q8_0 file is 209.4 GB, about 1.7 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
If memory is tight, the smallest file is the first one in the table at 0.5 GB, with the quality trade-offs that come with fewer bits.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Step 3.5 Flash
How much VRAM does Step 3.5 Flash need?
About 134.8 GB at Q4_K_M with an 8K context window. That is the 119.9 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run Step 3.5 Flash on an 8 GB GPU?
Yes, with the smaller files: the 0.5 GB file needs about 1.0 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Step 3.5 Flash GGUF?
It depends on the quant: file is 0.5 GB, IQ1_S is 41.3 GB, IQ1_M is 43.1 GB, IQ2_XXS is 48.0 GB, IQ2_XS is 55.2 GB, IQ2_S is 55.5 GB and 22 more.
What context length does Step 3.5 Flash support?
The GGUF header lists a trained context length of 262,144 tokens (256K).
Where can I download Step 3.5 Flash GGUF?
The files are on Hugging Face in the bartowski/stepfun-ai_Step-3.5-Flash-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Step 3.5 Flash use?
Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run Step 3.5 Flash on 8 GB VRAM?
The smallest quant for 8 GB VRAM is IQ1_S, which requires 46.8 GB memory at 8K context, so it is not feasible on 8 GB VRAM.
Does Step 3.5 Flash support vision tasks?
No, the model does not include vision capabilities.