Qwen3.8 Flash Next Heretic 2 needs about 4.1 GB of memory at Q4_K_M (a 2.8 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 9 quants, from 2.8 GB (Q4_K_M) to 188.2 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 3.3 GB.
Qwen3.8 Flash Next Heretic 2 – 177B parameter LLM with vision support
Qwen3.8 Flash Next Heretic 2 is a 177B parameter model based on the qwen4exp architecture. It supports vision and includes a mixture‑of‑experts component. The model is available in several GGUF quantizations for llama.cpp inference.
Good at
- Vision‑enabled text generation
- Large context handling up to 262144 tokens
- Mixture‑of‑experts for specialized tasks
Limits
- Requires a recent llama.cpp build with PR #27836 for MTP head
- Vision support needs mmproj file
- Large memory footprint for higher‑quality quants
Best for
- Vision‑centric content creation
- Long‑form document analysis
- Fine‑tuned expert inference
Who it’s for. Developers and researchers needing a high‑parameter, vision‑enabled LLM for specialized inference. The model is suitable for users who can manage the required hardware and software dependencies.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 2.8 GB (2 parts) | 4.1 GB | Fits | Get file → | |
| Q5_K_M | 2.8 GB (2 parts) | 4.1 GB | Fits | Get file → | |
| Q6_K | 2.8 GB (2 parts) | 4.1 GB | Fits | Get file → | |
| Q8_0 | 2.8 GB (2 parts) | 4.1 GB | Fits | Get file → | |
| Q8_0 | 4.1 GB | 5.5 GB | Fits | Get file → | |
| Q4_K_M | 140.4 GB (5 parts) | 141.7 GB | Too big | Get file → | |
| Q5_K_M | 151.7 GB (6 parts) | 153.1 GB | Too big | Get file → | |
| Q6_K | 168.7 GB (6 parts) | 170.0 GB | Too big | Get file → | |
| Q8_0 | 188.2 GB (6 parts) | 189.5 GB | Too big | Get file → |
For 8 GB VRAM, use the Q4_K_M or Q5_K_M quant (2.79 GB file, ~4.1 GB RAM at 8 k context). For 16 GB VRAM, Q8_0 (4.14 GB file, ~5.5 GB RAM) is a good balance. Larger quants (140–188 GB) require multi‑GPU or CPU inference.
What GPU do you need for Qwen3.8 Flash Next Heretic 2?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) 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: no quant fits at an 8K context.
- 6 GB VRAM: the best quant that fits is Q8_0 (4.1 GB file, about 5.5 GB in use).
- 8 GB VRAM: the best quant that fits is Q8_0 (4.1 GB file, about 5.5 GB in use).
- 12 GB VRAM: the best quant that fits is Q8_0 (4.1 GB file, about 5.5 GB in use).
- 16 GB VRAM: the best quant that fits is Q8_0 (4.1 GB file, about 5.5 GB in use).
- 24 GB VRAM: the best quant that fits is Q8_0 (4.1 GB file, about 5.5 GB in use).
- 32 GB VRAM: the best quant that fits is Q8_0 (4.1 GB file, about 5.5 GB in use).
- 48 GB VRAM: the best quant that fits is Q8_0 (4.1 GB file, about 5.5 GB in use).
Which Qwen3.8 Flash Next Heretic 2 quant should you choose?
Q4_K_M is the usual starting point for Qwen3.8 Flash Next Heretic 2: it balances file size and quality well. Here it is a 2.8 GB file.
The Q8_0 file is 2.8 GB, about 1.0 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.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Qwen3.8 Flash Next Heretic 2
How much VRAM does Qwen3.8 Flash Next Heretic 2 need?
About 4.1 GB at Q4_K_M with an 8K context window. That is the 2.8 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 Heretic 2 on an 8 GB GPU?
Yes, with Q8_0: the 4.1 GB file needs about 5.5 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 Heretic 2 GGUF?
It depends on the quant: Q4_K_M is 2.8 GB, Q5_K_M is 2.8 GB, Q6_K is 2.8 GB, Q8_0 is 2.8 GB, Q8_0 is 4.1 GB, Q4_K_M is 140.4 GB and 3 more.
What context length does Qwen3.8 Flash Next Heretic 2 support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.8 GB at 8K and 3.3 GB at 32K.
Where can I download Qwen3.8 Flash Next Heretic 2 GGUF?
The files are on Hugging Face in the spiritfather/Qwen3.8-Flash-Next-heretic-2-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 Heretic 2 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 Qwen3.8 Flash Next Heretic 2 on 8 GB VRAM?
Yes, using the Q4_K_M or Q5_K_M quant, which are 2.79 GB files and need about 4.1 GB RAM at 8 k context.
Does the model support vision?
Yes, vision is enabled via the mmproj file and requires the --mmproj flag in llama.cpp.
What llama.cpp version is needed?
A recent llama.cpp build with the qwen4exp merge and PR #27836 for the MTP head is required.