QW

Qwen3.8 27B Max Ablit Heretic

QwenGGUF by mradermacherUpdated 2026-10-14
Download on Hugging Face
Parameters27B
Context256K
Layers65
Architectureqwen35
Smallest quant0.0 GB
Quick answer

Qwen3.8 27B Max Ablit Heretic needs about 19.2 GB of memory at Q4_K_M (a 16.5 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 24 quants, from 0.0 GB to 22.1 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 8.7 GB.

Qwen3.8‑27B Max Ablit Heretic – a 27 B vision‑language model with 262 144‑token context

Qwen3.8‑27B Max Ablit Heretic is a 27 B parameter model built on the Qwen3.5 architecture. It supports vision‑language tasks and long‑horizon reasoning with a 262 144‑token context window. The model is released under the Apache‑2.0 license.

Good at

  • Vision‑language understanding
  • Long‑horizon agentic reasoning
  • Large context handling

Limits

  • Requires 8 GB+ VRAM for 8 k context
  • No native vision support in llama.cpp without mmproj
  • Large memory footprint for higher quants

Best for

  • Complex multi‑step reasoning tasks
  • Vision‑based document analysis
  • Long‑form content generation

Who it’s for. Developers and researchers needing a dense vision‑language model that can handle very long contexts. Suitable for local deployment with sufficient GPU memory.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
—0.0 GB 2.7 GB
Fits Get file →
IQ1_S7.1 GB 9.8 GB
Too big Get file →
IQ1_M7.6 GB 10.3 GB
Too big Get file →
IQ2_XXS8.4 GB 11.1 GB
Too big Get file →
IQ2_XS9.1 GB 11.8 GB
Too big Get file →
IQ2_S9.4 GB 12.0 GB
Too big Get file →
IQ2_M10.0 GB 12.7 GB
Too big Get file →
Q2_K_S10.2 GB 12.9 GB
Too big Get file →
Q2_K10.7 GB 13.4 GB
Too big Get file →
IQ3_XXS11.2 GB 13.9 GB
Too big Get file →
IQ3_XS12.0 GB 14.6 GB
Too big Get file →
Q3_K_S12.1 GB 14.8 GB
Too big Get file →
IQ3_S12.4 GB 15.1 GB
Too big Get file →
IQ3_M12.6 GB 15.3 GB
Too big Get file →
Q3_K_M13.3 GB 16.0 GB
Too big Get file →
Q3_K_L14.3 GB 17.0 GB
Too big Get file →
IQ4_XS15.1 GB 17.8 GB
Too big Get file →
Q4_015.5 GB 18.2 GB
Too big Get file →
Q4_K_S15.6 GB 18.3 GB
Too big Get file →
Q4_K_M16.5 GB 19.2 GB
Too big Get file →
Q4_117.1 GB 19.8 GB
Too big Get file →
Q5_K_S18.7 GB 21.4 GB
Too big Get file →
Q5_K_M19.2 GB 21.9 GB
Too big Get file →
Q6_K22.1 GB 24.8 GB
Too big Get file →

For 8 GB VRAM, use IQ1_S (7.15 GB) or IQ1_M (7.63 GB). If 12 GB is available, IQ2_S (9.09 GB) or Q2_K_S (10.25 GB) provide better quality. Higher quants such as Q4_K_M (16.55 GB) require 20 GB+ VRAM and are recommended only for high‑end GPUs.

What GPU do you need for Qwen3.8 27B Max Ablit Heretic?

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: the best quant that fits is the main file (0.0 GB file, about 2.7 GB in use).
  • 6 GB VRAM: the best quant that fits is the main file (0.0 GB file, about 2.7 GB in use).
  • 8 GB VRAM: the best quant that fits is the main file (0.0 GB file, about 2.7 GB in use).
  • 12 GB VRAM: the best quant that fits is IQ2_S (9.4 GB file, about 12.0 GB in use).
  • 16 GB VRAM: the best quant that fits is Q3_K_M (13.3 GB file, about 16.0 GB in use).
  • 24 GB VRAM: the best quant that fits is Q5_K_M (19.2 GB file, about 21.9 GB in use).
  • 32 GB VRAM: the best quant that fits is Q6_K (22.1 GB file, about 24.8 GB in use).

Which Qwen3.8 27B Max Ablit Heretic quant should you choose?

Q4_K_M is the usual starting point for Qwen3.8 27B Max Ablit Heretic: it balances file size and quality well. Here it is a 16.5 GB file.

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

Architectureqwen35
Parameters27B
Layers65
Trained context262,144 tokens
KV cache (fp16)260 KB per token

Source

Quantized bymradermacher
FamilyQwen
Licenseapache-2.0
Downloads (HF)6,974
Last updated2026-10-14

Frequently asked questions about Qwen3.8 27B Max Ablit Heretic

How much VRAM does Qwen3.8 27B Max Ablit Heretic need?

About 19.2 GB at Q4_K_M with an 8K context window. That is the 16.5 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 27B Max Ablit Heretic on an 8 GB GPU?

Yes, with the smaller files: the 0.0 GB file needs about 2.7 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Qwen3.8 27B Max Ablit Heretic GGUF?

It depends on the quant: file is 0.0 GB, IQ1_S is 7.1 GB, IQ1_M is 7.6 GB, IQ2_XXS is 8.4 GB, IQ2_XS is 9.1 GB, IQ2_S is 9.4 GB and 18 more.

What context length does Qwen3.8 27B Max Ablit Heretic support?

The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 2.2 GB at 8K and 8.7 GB at 32K.

Where can I download Qwen3.8 27B Max Ablit Heretic GGUF?

The files are on Hugging Face in the mradermacher/Qwen3.8-27B-Max-Ablit-Heretic-i1-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 27B Max Ablit Heretic 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 this model on 8 GB VRAM?

Yes, with IQ1_S or IQ1_M quants, which use about 9–10 GB at 8 k context.

Does this model support vision tasks in llama.cpp?

Vision support requires mmproj files; the GGUF files are for language inference only.

What license applies to this model?

The model is released under the Apache‑2.0 license.

Other GGUFs of Qwen3.8-27B-Max-Ablit-Heretic

Similar models

Browse all →

More from mradermacher

All by mradermacher →

More Qwen models

All Qwen →

Other medium models

All medium models →