Qwen3.8 Flash Next needs about 120.9 GB of memory at Q4_K_M (a 119.6 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 32 quants, from 0.6 GB to 354.0 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 3.2 GB.
Qwen3.8 Flash Next – 177B parameters, 262,144 token context, multimodal
Qwen3.8 Flash Next is a 177B‑parameter model with a 262,144‑token context window. It supports text and image inputs and is built on the qwen4exp architecture. The model is available in multiple GGUF quantizations for llama.cpp inference.
Good at
- Large context handling
- Multimodal text‑image tasks
- Efficient inference with low‑bit quantizations
Limits
- No vision support without mmproj
- Requires recent llama.cpp release
- Large memory for high‑bit quantizations
Best for
- Long‑form content generation
- Image‑based question answering
- Fine‑tuning on specialized datasets
Who it’s for. Developers and researchers needing a high‑capacity multimodal model for inference or fine‑tuning. Suitable for applications that can accommodate the model’s large memory footprint and require extensive context handling.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.6 GB | 1.9 GB | Fits | Get file → | |
| Q4_0 | 2.2 GB | 3.5 GB | Fits | Get file → | |
| Q8_0 | 4.1 GB | 5.4 GB | Fits | Get file → | |
| BF16 | 7.8 GB | 9.1 GB | Too big | Get file → | |
| IQ1_S | 70.1 GB (2 parts) | 71.4 GB | Too big | Get file → | |
| IQ1_M | 72.0 GB (2 parts) | 73.3 GB | Too big | Get file → | |
| IQ2_XXS | 75.2 GB (2 parts) | 76.5 GB | Too big | Get file → | |
| IQ2_XS | 77.7 GB (2 parts) | 79.0 GB | Too big | Get file → | |
| IQ2_S | 77.8 GB (2 parts) | 79.1 GB | Too big | Get file → | |
| IQ2_M | 80.4 GB (3 parts) | 81.7 GB | Too big | Get file → | |
| Q2_K | 80.9 GB (3 parts) | 82.2 GB | Too big | Get file → | |
| IQ3_XXS | 88.0 GB (3 parts) | 89.3 GB | Too big | Get file → | |
| Q3_K_S | 89.4 GB (3 parts) | 90.7 GB | Too big | Get file → | |
| IQ3_XS | 91.9 GB (3 parts) | 93.3 GB | Too big | Get file → | |
| Q3_K_M | 92.0 GB (3 parts) | 93.3 GB | Too big | Get file → | |
| Q3_K_L | 93.2 GB (3 parts) | 94.5 GB | Too big | Get file → | |
| IQ3_M | 93.2 GB (3 parts) | 94.5 GB | Too big | Get file → | |
| IQ4_XS | 97.7 GB (3 parts) | 99.0 GB | Too big | Get file → | |
| IQ4_NL | 100.3 GB (3 parts) | 101.6 GB | Too big | Get file → | |
| Q4_0 | 100.6 GB (3 parts) | 101.9 GB | Too big | Get file → | |
| Q2_K_L | 107.1 GB (4 parts) | 108.5 GB | Too big | Get file → | |
| Q4_1 | 111.3 GB (3 parts) | 112.6 GB | Too big | Get file → | |
| Q4_K_S | 113.1 GB (3 parts) | 114.4 GB | Too big | Get file → | |
| Q3_K_XL | 119.3 GB (4 parts) | 120.7 GB | Too big | Get file → | |
| Q4_K_M | 119.6 GB (4 parts) | 120.9 GB | Too big | Get file → | |
| Q5_K_S | 128.1 GB (4 parts) | 129.4 GB | Too big | Get file → | |
| Q5_K_M | 134.7 GB (4 parts) | 136.0 GB | Too big | Get file → | |
| Q4_K_L | 139.3 GB (5 parts) | 140.6 GB | Too big | Get file → | |
| Q5_K_L | 151.1 GB (5 parts) | 152.4 GB | Too big | Get file → | |
| Q6_K | 168.1 GB (5 parts) | 169.4 GB | Too big | Get file → | |
| Q8_0 | 188.3 GB (6 parts) | 189.6 GB | Too big | Get file → | |
| BF16 | 354.0 GB (9 parts) | 355.3 GB | Too big | Get file → |
For 8‑GB VRAM, use Q4_0 (2.2 GB file, 3.5 GB memory at 8k context). For 16‑GB VRAM, Q4_K_M (119.6 GB file, 120.9 GB memory) offers a balance of size and performance. For 32‑GB VRAM, Q5_K_L (151.06 GB file, 152.4 GB memory) provides higher fidelity.
What GPU do you need for Qwen3.8 Flash Next?
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 Q4_0 (2.2 GB file, about 3.5 GB in use).
- 6 GB VRAM: the best quant that fits is Q8_0 (4.1 GB file, about 5.4 GB in use).
- 8 GB VRAM: the best quant that fits is Q8_0 (4.1 GB file, about 5.4 GB in use).
- 12 GB VRAM: the best quant that fits is BF16 (7.8 GB file, about 9.1 GB in use).
- 16 GB VRAM: the best quant that fits is BF16 (7.8 GB file, about 9.1 GB in use).
- 24 GB VRAM: the best quant that fits is BF16 (7.8 GB file, about 9.1 GB in use).
- 32 GB VRAM: the best quant that fits is BF16 (7.8 GB file, about 9.1 GB in use).
- 48 GB VRAM: the best quant that fits is BF16 (7.8 GB file, about 9.1 GB in use).
Which Qwen3.8 Flash Next quant should you choose?
Q4_K_M is the usual starting point for Qwen3.8 Flash Next: it balances file size and quality well. Here it is a 119.6 GB file.
The Q8_0 file is 4.1 GB, about 0.0 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 0.6 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 Qwen3.8 Flash Next
How much VRAM does Qwen3.8 Flash Next need?
About 120.9 GB at Q4_K_M with an 8K context window. That is the 119.6 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 Qwen3.8 Flash Next on an 8 GB GPU?
Yes, with Q8_0: the 4.1 GB file needs about 5.4 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Qwen3.8 Flash Next GGUF?
It depends on the quant: file is 0.6 GB, Q4_0 is 2.2 GB, Q8_0 is 4.1 GB, BF16 is 7.8 GB, IQ1_S is 70.1 GB, IQ1_M is 72.0 GB and 26 more.
What context length does Qwen3.8 Flash Next support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.8 GB at 8K and 3.2 GB at 32K.
Where can I download Qwen3.8 Flash Next GGUF?
The files are on Hugging Face in the bartowski/Qwen3.8-Flash-Next-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 Flash Next 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 Qwen3.8 Flash Next on 8 GB VRAM?
Use the Q4_0 quantization; it requires 3.5 GB memory at 8k context, which fits within 8 GB VRAM with headroom.
Does the model support image input?
Yes, but you need the mmproj file. The mmproj is downloaded automatically with the -hf flag or passed with --mmproj.
What llama.cpp release is required?
The quants were made with llama.cpp release b10665; newer releases are required for full compatibility.