Qwen3.8 Flash Next S abliterated Strata needs about 2.8 GB of memory at its main file (a 2.1 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 5 quants, from 0.0 GB to 42.3 GB.
Qwen3.8 Flash Next S abliterated Strata – 2‑bit trellis build for Strata
This model is a 2‑bit trellis version of Qwen3.8‑Flash‑Next, optimized for Strata. It runs on one or two RTX 3090 GPUs with experts in system RAM. The backbone and head are packed as trellis, leaving more VRAM for experts.
Good at
- Efficient long‑context inference
- Low‑bit quantization for memory‑constrained GPUs
- Strata‑compatible deployment
Limits
- Requires Linux and NVIDIA GPU
- Needs ~64 GB RAM for S build
- No vision support in this build
Best for
- Running on RTX 3090 with Strata
- Deploying large‑context models on limited VRAM
- Testing low‑bit quantization performance
Who it’s for. Developers and researchers who want to experiment with Qwen3.8‑Flash‑Next in a Strata environment. The model is suitable for those who have access to high‑VRAM GPUs and sufficient system RAM.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.0 GB | 0.5 GB | Fits | Get file → | |
| — | 2.1 GB | 2.8 GB | Fits | Get file → | |
| — | 2.1 GB | 2.8 GB | Fits | Get file → | |
| — | 34.4 GB | 39.0 GB | Too big | Get file → | |
| — | 42.3 GB | 47.9 GB | Too big | Get file → |
The 2.08 GB file uses 2.8 GB of VRAM at 8 k context and is suitable for a single RTX 3090. The 34.39 GB file needs 39 GB VRAM and is best for dual‑GPU setups. The 42.3 GB file requires 47.9 GB VRAM and is recommended for the largest context windows.
What GPU do you need for Qwen3.8 Flash Next S abliterated Strata?
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 (2.1 GB file, about 2.8 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (2.1 GB file, about 2.8 GB in use).
- 8 GB VRAM: the best quant that fits is the main file (2.1 GB file, about 2.8 GB in use).
- 12 GB VRAM: the best quant that fits is the main file (2.1 GB file, about 2.8 GB in use).
- 16 GB VRAM: the best quant that fits is the main file (2.1 GB file, about 2.8 GB in use).
- 24 GB VRAM: the best quant that fits is the main file (2.1 GB file, about 2.8 GB in use).
- 32 GB VRAM: the best quant that fits is the main file (2.1 GB file, about 2.8 GB in use).
- 48 GB VRAM: the best quant that fits is the main file (42.3 GB file, about 47.9 GB in use).
Which Qwen3.8 Flash Next S abliterated Strata quant should you choose?
If memory is tight, the smallest file is the first one in the table at 0.0 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 Qwen3.8 Flash Next S abliterated Strata
How much VRAM does Qwen3.8 Flash Next S abliterated Strata need?
About 2.8 GB at its main file with an 8K context window. That is the 2.1 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 Qwen3.8 Flash Next S abliterated Strata on an 8 GB GPU?
Yes, with the smaller files: the 2.1 GB file needs about 2.8 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Qwen3.8 Flash Next S abliterated Strata GGUF?
It depends on the quant: file is 0.0 GB, file is 2.1 GB, file is 2.1 GB, file is 34.4 GB, file is 42.3 GB.
Where can I download Qwen3.8 Flash Next S abliterated Strata GGUF?
The files are on Hugging Face in the alesha-pro/Qwen3.8-Flash-Next-S-abliterated-Strata-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 Qwen3.8 Flash Next S abliterated Strata use?
Hugging Face lists the license as other. 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 the 2.08 GB quant on a single RTX 3090?
Yes, the 2.08 GB quant uses about 2.8 GB VRAM at 8 k context, which fits comfortably on a single RTX 3090.
What RAM is needed for the S build?
The S build requires about 64 GB of system RAM, with 48 GB for the smaller S‑small variant.
Does this model support vision?
No, the Strata build runs with vision disabled, saving about 1.3 GB of VRAM.