Qwen Qwen3 VL 32B Instruct Heretic needs about 22.4 GB of memory at Q4_K_M (a 19.8 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 3 quants, from 19.8 GB (Q4_K_M) to 34.8 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 8.6 GB.
Qwen Qwen3 VL 32B Instruct Heretic – a vision‑language model in GGUF format
Qwen Qwen3 VL 32B Instruct Heretic is a 32.8B parameter model with vision capabilities. It is a quantized version of Qwen3-VL-32B-Instruct, converted to GGUF by ThalisAI. The model supports a 262144 token context window and is available in Q4_K_M, Q6_K, and Q8_0 quantizations.
Good at
- Vision‑language tasks
- Long‑context reasoning
- Multimodal generation
Limits
- Vision only; image features crash in Ollama
- Requires separate mmproj file
- Large memory footprint for 32B model
Best for
- Image‑to‑text summarization
- Visual question answering
- Multimodal chatbot
Who it’s for. Developers building vision‑enabled LLM applications who need a large context window and are comfortable using llama.cpp. The model is suitable for research and production environments that can allocate sufficient VRAM.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 19.8 GB | 22.4 GB | Too big | Get file → | |
| Q6_K | 26.9 GB | 29.5 GB | Too big | Get file → | |
| Q8_0 | 34.8 GB | 37.5 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_M (file size 19.76 GB, estimated 22.4 GB memory at 8 k context). For 12 GB VRAM, Q6_K (26.88 GB, 29.5 GB). For 16 GB or more, Q8_0 (34.82 GB, 37.5 GB). Memory estimates are approximate.
What GPU do you need for Qwen Qwen3 VL 32B Instruct 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: 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: no quant fits at an 8K context.
- 16 GB VRAM: no quant fits at an 8K context.
- 24 GB VRAM: the best quant that fits is Q4_K_M (19.8 GB file, about 22.4 GB in use).
- 32 GB VRAM: the best quant that fits is Q6_K (26.9 GB file, about 29.5 GB in use).
- 48 GB VRAM: the best quant that fits is Q8_0 (34.8 GB file, about 37.5 GB in use).
Which Qwen Qwen3 VL 32B Instruct Heretic quant should you choose?
Q4_K_M is the usual starting point for Qwen Qwen3 VL 32B Instruct Heretic: it balances file size and quality well. Here it is a 19.8 GB file.
The Q8_0 file is 34.8 GB, about 1.8 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 Qwen Qwen3 VL 32B Instruct Heretic
How much VRAM does Qwen Qwen3 VL 32B Instruct Heretic need?
About 22.4 GB at Q4_K_M with an 8K context window. That is the 19.8 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 Qwen Qwen3 VL 32B Instruct Heretic on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 22.4 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Qwen Qwen3 VL 32B Instruct Heretic GGUF?
It depends on the quant: Q4_K_M is 19.8 GB, Q6_K is 26.9 GB, Q8_0 is 34.8 GB.
What context length does Qwen Qwen3 VL 32B Instruct 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 2.1 GB at 8K and 8.6 GB at 32K.
Where can I download Qwen Qwen3 VL 32B Instruct Heretic GGUF?
The files are on Hugging Face in the ThalisAI/Qwen3-VL-32B-Instruct-heretic repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Qwen Qwen3 VL 32B Instruct Heretic 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 this model on 8 GB VRAM?
Yes, using the Q4_K_M quantization. It requires about 22.4 GB of VRAM at 8 k context, so 8 GB is insufficient; you need at least 22.4 GB.
Does this model support text‑only inference?
Ollama can load the model for text‑only chat, but vision features will crash due to a known bug. Use llama.cpp for full functionality.
Where do I get the multimodal projector file?
Download the official mmproj file from the Qwen/Qwen3-VL-32B-Instruct-GGUF repository.