LongCat Flash Lite Uncensored Heretic Native MTP Preserved needs about 41.3 GB of memory at Q4_K_M (a 40.3 GB file) with an 8K context window, so it fits on a 48 GB GPU. This repository has 11 quants, from 24.2 GB (Q2_K) to 138.2 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 2.1 GB.
69.1B parameter Mixture-of-Experts model with 256k context length
LongCat Flash Lite Uncensored Heretic Native MTP Preserved is a 69.1B parameter Mixture-of-Experts model. It supports a 256k token context length via the YaRN method and is available in multiple GGUF quantizations. The model is released under the MIT license.
Good at
- Large context inference
- Uncensored text generation
- Efficient MoE usage
Limits
- Requires significant VRAM for higher quantizations
- No vision support
- Large context may need more RAM
Best for
- Long‑form content creation
- Agentic task planning
- Coding assistance
Who it’s for. Developers and researchers who need a large‑scale, uncensored language model with extended context. The model is suitable for local deployment using llama.cpp.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_K | 24.2 GB | 25.2 GB | Too big | Get file → | |
| Q3_K_S | 29.9 GB | 30.9 GB | Too big | Get file → | |
| Q3_K_M | 31.6 GB | 32.6 GB | Too big | Get file → | |
| Q3_K_L | 33.0 GB | 34.0 GB | Too big | Get file → | |
| Q4_K_S | 39.2 GB | 40.2 GB | Too big | Get file → | |
| Q4_K_M | 40.3 GB | 41.3 GB | Too big | Get file → | |
| Q5_K_S | 47.6 GB | 48.6 GB | Too big | Get file → | |
| Q5_K_M | 48.3 GB | 49.3 GB | Too big | Get file → | |
| Q6_K | 56.7 GB | 57.7 GB | Too big | Get file → | |
| Q8_0 | 73.4 GB | 74.4 GB | Too big | Get file → | |
| BF16 | 138.2 GB | 139.2 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_S (39.21 GB file, ~40.2 GB memory at 8k context). For 16 GB VRAM, Q5_K_S (47.58 GB file, ~48.6 GB memory) or Q4_K_M (40.31 GB file, ~41.3 GB memory). Higher quantizations require more memory and are not recommended for limited GPU budgets.
What GPU do you need for LongCat Flash Lite Uncensored Heretic Native MTP Preserved?
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: no quant fits at an 8K context.
- 32 GB VRAM: the best quant that fits is Q3_K_S (29.9 GB file, about 30.9 GB in use).
- 48 GB VRAM: the best quant that fits is Q4_K_M (40.3 GB file, about 41.3 GB in use).
Which LongCat Flash Lite Uncensored Heretic Native MTP Preserved quant should you choose?
Q4_K_M is the usual starting point for LongCat Flash Lite Uncensored Heretic Native MTP Preserved: it balances file size and quality well. Here it is a 40.3 GB file.
The Q8_0 file is 73.4 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 Q2_K at 24.2 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Frequently asked questions about LongCat Flash Lite Uncensored Heretic Native MTP Preserved
How much VRAM does LongCat Flash Lite Uncensored Heretic Native MTP Preserved need?
About 41.3 GB at Q4_K_M with an 8K context window. That is the 40.3 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 LongCat Flash Lite Uncensored Heretic Native MTP Preserved on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 25.2 GB, so you would need a larger GPU or CPU offloading.
What is the file size of LongCat Flash Lite Uncensored Heretic Native MTP Preserved GGUF?
It depends on the quant: Q2_K is 24.2 GB, Q3_K_S is 29.9 GB, Q3_K_M is 31.6 GB, Q3_K_L is 33.0 GB, Q4_K_S is 39.2 GB, Q4_K_M is 40.3 GB and 5 more.
What context length does LongCat Flash Lite Uncensored Heretic Native MTP Preserved support?
The GGUF header lists a trained context length of 327,680 tokens (320K). The KV cache grows with the context you use: about 0.5 GB at 8K and 2.1 GB at 32K.
Where can I download LongCat Flash Lite Uncensored Heretic Native MTP Preserved GGUF?
The files are on Hugging Face in the llmfan46/LongCat-Flash-Lite-Uncensored-Heretic-Native-MTP-Preserved-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 LongCat Flash Lite Uncensored Heretic Native MTP Preserved use?
Hugging Face lists the license as mit. 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 Q4_K_S quantization; it requires about 40.2 GB memory at 8k context, which is beyond 8 GB VRAM, so it is not feasible.
Does this model support vision tasks?
No, the model has no vision capabilities.
What license applies to this model?
The model is released under the MIT license.