Qwen3-Vl-30B-A3B-Thinking-1M needs about 19.9 GB of memory at Q4_K_M (a 18.6 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 27 quants, from 8.2 GB (TQ1_0) to 61.1 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 3.2 GB.
30B vision‑language model with 1M‑token context and MoE architecture
Qwen3‑VL‑30B‑A3B‑Thinking‑1M is a 30 billion‑parameter model with 3 billion active MoE parameters. It supports vision, reasoning, and long‑context text up to 1 million tokens. The model is released under Apache‑2.0 and is available in multiple GGUF quantizations.
Good at
- Vision‑language reasoning
- Long‑context text generation
- Multimodal agent tasks
Limits
- Requires large VRAM for full precision
- No native vision support in llama.cpp without quant
- Limited to 8 k context in most quants
Best for
- Analyzing complex images with text prompts
- Generating code from screenshots
- Extended document summarization
Who it’s for. Developers building multimodal applications that need deep visual understanding and long‑term memory. Researchers testing vision‑language reasoning at scale.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| TQ1_0 | 8.2 GB | 9.5 GB | Too big | Get file → | |
| IQ1_S | 9.1 GB | 10.4 GB | Too big | Get file → | |
| IQ1_M | 9.7 GB | 11.0 GB | Too big | Get file → | |
| IQ2_XXS | 10.4 GB | 11.7 GB | Too big | Get file → | |
| IQ2_M | 10.9 GB | 12.2 GB | Too big | Get file → | |
| Q2_K | 11.3 GB | 12.6 GB | Too big | Get file → | |
| Q2_K_L | 11.3 GB | 12.6 GB | Too big | Get file → | |
| Q2_K_XL | 11.8 GB | 13.1 GB | Too big | Get file → | |
| IQ3_XXS | 12.9 GB | 14.2 GB | Too big | Get file → | |
| Q3_K_S | 13.3 GB | 14.6 GB | Too big | Get file → | |
| Q3_K_XL | 13.9 GB | 15.2 GB | Too big | Get file → | |
| Q3_K_M | 14.7 GB | 16.0 GB | Too big | Get file → | |
| IQ4_XS | 16.4 GB | 17.7 GB | Too big | Get file → | |
| IQ4_NL | 17.3 GB | 18.6 GB | Too big | Get file → | |
| Q4_0 | 17.4 GB | 18.7 GB | Too big | Get file → | |
| Q4_K_S | 17.5 GB | 18.8 GB | Too big | Get file → | |
| Q4_K_XL | 17.7 GB | 19.0 GB | Too big | Get file → | |
| Q4_K_M | 18.6 GB | 19.9 GB | Too big | Get file → | |
| Q4_1 | 19.2 GB | 20.5 GB | Too big | Get file → | |
| Q5_K_S | 21.1 GB | 22.4 GB | Too big | Get file → | |
| Q5_K_M | 21.7 GB | 23.0 GB | Too big | Get file → | |
| Q5_K_XL | 21.7 GB | 23.0 GB | Too big | Get file → | |
| Q6_K | 25.1 GB | 26.4 GB | Too big | Get file → | |
| Q6_K_XL | 26.3 GB | 27.6 GB | Too big | Get file → | |
| Q8_0 | 32.5 GB | 33.8 GB | Too big | Get file → | |
| Q8_K_XL | 36.0 GB | 37.3 GB | Too big | Get file → | |
| BF16 | 61.1 GB (2 parts) | 62.4 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_S (17.46 GB file, ~18.8 GB RAM). For 12 GB VRAM, Q4_K_M (18.56 GB file, ~19.9 GB RAM). For 16 GB VRAM, Q5_K_S (21.08 GB file, ~22.4 GB RAM). Estimates are approximate.
What GPU do you need for Qwen3-Vl-30B-A3B-Thinking-1M?
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 IQ2_XXS (10.4 GB file, about 11.7 GB in use).
- 16 GB VRAM: the best quant that fits is Q3_K_M (14.7 GB file, about 16.0 GB in use).
- 24 GB VRAM: the best quant that fits is Q5_K_XL (21.7 GB file, about 23.0 GB in use).
- 32 GB VRAM: the best quant that fits is Q6_K_XL (26.3 GB file, about 27.6 GB in use).
- 48 GB VRAM: the best quant that fits is Q8_K_XL (36.0 GB file, about 37.3 GB in use).
Which Qwen3-Vl-30B-A3B-Thinking-1M quant should you choose?
Q4_K_M is the usual starting point for Qwen3-Vl-30B-A3B-Thinking-1M: it balances file size and quality well. Here it is a 18.6 GB file.
The Q8_0 file is 32.5 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.
If memory is tight, the smallest file is TQ1_0 at 8.2 GB, with the quality trade-offs that come with fewer bits. Lower bit counts mean smaller files and more quality loss.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Qwen3-Vl-30B-A3B-Thinking-1M
How much VRAM does Qwen3-Vl-30B-A3B-Thinking-1M need?
About 19.9 GB at Q4_K_M with an 8K context window. That is the 18.6 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-Vl-30B-A3B-Thinking-1M on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 9.5 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Qwen3-Vl-30B-A3B-Thinking-1M GGUF?
It depends on the quant: TQ1_0 is 8.2 GB, IQ1_S is 9.1 GB, IQ1_M is 9.7 GB, IQ2_XXS is 10.4 GB, IQ2_M is 10.9 GB, Q2_K is 11.3 GB and 21 more.
What context length does Qwen3-Vl-30B-A3B-Thinking-1M support?
The GGUF header lists a trained context length of 1,000,000 tokens (977K). The KV cache grows with the context you use: about 0.8 GB at 8K and 3.2 GB at 32K.
Where can I download Qwen3-Vl-30B-A3B-Thinking-1M GGUF?
The files are on Hugging Face in the unsloth/Qwen3-VL-30B-A3B-Thinking-1M-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-Vl-30B-A3B-Thinking-1M 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 Qwen3‑VL‑30B‑A3B‑Thinking‑1M on 8 GB VRAM?
Only with a low‑precision quant like Q4_K_S, which requires about 18.8 GB of RAM; 8 GB VRAM is insufficient.
Does the model support vision in llama.cpp?
Yes, but only with the provided GGUF files; the base model requires a quantized version for llama.cpp.
What context length can I use with the Q4_K_S quant?
The estimated memory at 8 k context is 18.8 GB; larger contexts will increase memory usage proportionally.