Holo4 35B A3B Heretic GSQ RCO needs about 14.3 GB of memory at IQ3_XXS (a 13.1 GB file) with an 8K context window, so it fits on a 16 GB GPU. This repository has 3 quants, from 0.9 GB (Q8_0) to 15.2 GB (IQ3_S). Longer context needs more: at 32K tokens the KV cache alone adds about 2.8 GB.
Holo4‑35B‑A3B Heretic GSQ RCO – a 35B vision‑language Mixture‑of‑Experts model
Holo4‑35B‑A3B is a ~35B Mixture‑of‑Experts vision‑language model built on Qwen3.6‑MoE. It supports reasoning, tools, and vision, and is designed for computer‑use tasks behind an agent harness. The model offers a 262,144‑token context window and is released under Apache‑2.0.
Good at
- Vision‑language reasoning
- Tool‑based interaction
- Large context handling
Limits
- No native vision support in llama.cpp
- Requires 8k context for memory estimates
- Large VRAM needed for higher quantizations
Best for
- Automated UI testing
- Code generation with tool calls
- Complex document analysis
Who it’s for. Developers building agent systems that need vision and tool integration. Researchers exploring large‑scale VLMs with Mixture‑of‑Experts architecture.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q8_0 | 0.9 GB | 2.1 GB | Fits | Get file → | |
| IQ3_XXS | 13.1 GB | 14.3 GB | Too big | Get file → | |
| IQ3_S | 15.2 GB | 16.4 GB | Too big | Get file → |
For 8‑GB VRAM, use Q8_0 (0.91 GB file, 2.1 GB memory at 8k context). If more VRAM is available, IQ3_XXS (13.07 GB file, 14.3 GB memory) or IQ3_S (15.23 GB file, 16.4 GB memory) provide higher compression but require more memory.
What GPU do you need for Holo4 35B A3B Heretic GSQ RCO?
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 Q8_0 (0.9 GB file, about 2.1 GB in use).
- 6 GB VRAM: the best quant that fits is Q8_0 (0.9 GB file, about 2.1 GB in use).
- 8 GB VRAM: the best quant that fits is Q8_0 (0.9 GB file, about 2.1 GB in use).
- 12 GB VRAM: the best quant that fits is Q8_0 (0.9 GB file, about 2.1 GB in use).
- 16 GB VRAM: the best quant that fits is IQ3_XXS (13.1 GB file, about 14.3 GB in use).
- 24 GB VRAM: the best quant that fits is IQ3_S (15.2 GB file, about 16.4 GB in use).
Which Holo4 35B A3B Heretic GSQ RCO quant should you choose?
IQ3_XXS (13.1 GB) is the middle option in this repository. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.
If memory is tight, the smallest file is Q8_0 at 0.9 GB, with the quality trade-offs that come with fewer bits. Close to lossless. It costs roughly twice the memory of a 4-bit file, so it suits cases where you have plenty of VRAM.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Holo4 35B A3B Heretic GSQ RCO
How much VRAM does Holo4 35B A3B Heretic GSQ RCO need?
About 14.3 GB at IQ3_XXS with an 8K context window. That is the 13.1 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 Holo4 35B A3B Heretic GSQ RCO on an 8 GB GPU?
Yes, with Q8_0: the 0.9 GB file needs about 2.1 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Holo4 35B A3B Heretic GSQ RCO GGUF?
It depends on the quant: Q8_0 is 0.9 GB, IQ3_XXS is 13.1 GB, IQ3_S is 15.2 GB.
What context length does Holo4 35B A3B Heretic GSQ RCO support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.7 GB at 8K and 2.8 GB at 32K.
Where can I download Holo4 35B A3B Heretic GSQ RCO GGUF?
The files are on Hugging Face in the lavendercomputing/Holo4-35B-A3B-Heretic-GSQ-RCO-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 Holo4 35B A3B Heretic GSQ RCO 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, the Q8_0 quantization uses about 2.1 GB of VRAM at 8k context, fitting within 8 GB.
Does this model support vision in llama.cpp?
No, llama.cpp does not natively support vision for this model; it requires an external harness to provide screenshots.
What license applies to this model?
The model is released under Apache‑2.0, as stated in the Hugging Face repository.