G4 Dark Soul 26B A4B needs about 25.6 GB of memory at Q4_K_M (a 17.0 GB file) with an 8K context window, so it fits on a 32 GB GPU. This repository has 27 quants, from 0.1 GB to 50.5 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 32.2 GB.
26B gemma4 model with vision and roleplay abilities
G4 Dark Soul 26B A4B is a 26‑billion parameter gemma4 model. It supports text, image, and audio input and is optimized for roleplay, creative writing, and storytelling. The model is released under Apache‑2.0 and can be run with llama.cpp or compatible tools.
Good at
- Roleplay and character dialogue
- Creative writing and story generation
- Vision tasks with multimodal input
Limits
- Large context size requires high VRAM
- No vision support without mmproj file
- Requires recent llama.cpp release
Best for
- Developing interactive fiction engines
- Generating detailed narrative scenes
- Creating multimodal content with images
Who it’s for. Developers and writers looking for a large multimodal model for roleplay and creative writing. The model is suitable for projects that can accommodate its memory demands and need vision capabilities.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.1 GB | 8.6 GB | Too big | Get file → | |
| IQ2_XS | 10.1 GB | 18.7 GB | Too big | Get file → | |
| IQ2_S | 10.2 GB | 18.8 GB | Too big | Get file → | |
| IQ2_M | 10.7 GB | 19.3 GB | Too big | Get file → | |
| Q2_K | 11.0 GB | 19.5 GB | Too big | Get file → | |
| Q2_K_L | 11.1 GB | 19.7 GB | Too big | Get file → | |
| IQ3_XXS | 12.2 GB | 20.7 GB | Too big | Get file → | |
| IQ3_XS | 12.4 GB | 21.0 GB | Too big | Get file → | |
| Q3_K_S | 12.6 GB | 21.1 GB | Too big | Get file → | |
| Q3_K_M | 13.0 GB | 21.6 GB | Too big | Get file → | |
| Q3_K_L | 13.2 GB | 21.8 GB | Too big | Get file → | |
| IQ3_M | 13.3 GB | 21.8 GB | Too big | Get file → | |
| Q3_K_XL | 13.4 GB | 21.9 GB | Too big | Get file → | |
| IQ4_XS | 14.2 GB | 22.8 GB | Too big | Get file → | |
| IQ4_NL | 14.7 GB | 23.3 GB | Too big | Get file → | |
| Q4_0 | 14.8 GB | 23.3 GB | Too big | Get file → | |
| Q4_K_S | 15.8 GB | 24.4 GB | Too big | Get file → | |
| Q4_1 | 16.1 GB | 24.7 GB | Too big | Get file → | |
| Q4_K_M | 17.0 GB | 25.6 GB | Too big | Get file → | |
| Q4_K_L | 17.2 GB | 25.8 GB | Too big | Get file → | |
| Q5_K_S | 18.1 GB | 26.7 GB | Too big | Get file → | |
| Q5_K_M | 19.3 GB | 27.9 GB | Too big | Get file → | |
| Q5_K_L | 19.5 GB | 28.1 GB | Too big | Get file → | |
| Q6_K | 22.9 GB | 31.4 GB | Too big | Get file → | |
| Q6_K_L | 23.0 GB | 31.6 GB | Too big | Get file → | |
| Q8_0 | 26.9 GB | 35.4 GB | Too big | Get file → | |
| BF16 | 50.5 GB | 59.1 GB | Too big | Get file → |
For 8‑GB VRAM, use IQ2_XS (10.13 GB) or IQ2_S (10.23 GB). For 16‑GB VRAM, Q4_K_M (17.04 GB) balances size and performance. For 24‑GB VRAM, Q5_K_M (19.32 GB) offers higher quality. Estimates are approximate.
What GPU do you need for G4 Dark Soul 26B A4B?
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: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: the best quant that fits is the main file (0.1 GB file, about 8.6 GB in use).
- 16 GB VRAM: the best quant that fits is the main file (0.1 GB file, about 8.6 GB in use).
- 24 GB VRAM: the best quant that fits is Q4_0 (14.8 GB file, about 23.3 GB in use).
- 32 GB VRAM: the best quant that fits is Q6_K_L (23.0 GB file, about 31.6 GB in use).
- 48 GB VRAM: the best quant that fits is Q8_0 (26.9 GB file, about 35.4 GB in use).
Which G4 Dark Soul 26B A4B quant should you choose?
Q4_K_M is the usual starting point for G4 Dark Soul 26B A4B: it balances file size and quality well. Here it is a 17.0 GB file.
The Q8_0 file is 26.9 GB, about 1.6 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 0.1 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 G4 Dark Soul 26B A4B
How much VRAM does G4 Dark Soul 26B A4B need?
About 25.6 GB at Q4_K_M with an 8K context window. That is the 17.0 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 G4 Dark Soul 26B A4B on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 8.6 GB, so you would need a larger GPU or CPU offloading.
What is the file size of G4 Dark Soul 26B A4B GGUF?
It depends on the quant: file is 0.1 GB, IQ2_XS is 10.1 GB, IQ2_S is 10.2 GB, IQ2_M is 10.7 GB, Q2_K is 11.0 GB, Q2_K_L is 11.1 GB and 21 more.
What context length does G4 Dark Soul 26B A4B support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 8.1 GB at 8K and 32.2 GB at 32K.
Where can I download G4 Dark Soul 26B A4B GGUF?
The files are on Hugging Face in the bartowski/Vortex5_G4-Dark-Soul-26B-A4B-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 G4 Dark Soul 26B A4B 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?
Use the IQ2_XS or IQ2_S quantizations; they require about 18–19 GB of VRAM at 8k context, so 8 GB is insufficient.
Does this model support vision?
Yes, it supports image and audio input when paired with the mmproj projector files included in the repo.
What license applies to this model?
The model is released under the Apache‑2.0 license.