DeepSeek V4 Flash 0731 REAP 150b TQ3 4S needs about 0.8 GB of memory at its main file (a 0.2 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 2 quants, from 0.2 GB to 65.5 GB.
DeepSeek V4 Flash 150B with selective TQ3_4S quant, 256K context
DeepSeek V4 Flash 150B REAP variant is a 150‑billion‑parameter model with routed experts pruned to 132 per layer. The GGUF file is hand‑tuned with TQ3_4S quantization, enabling 256K context length. It is released under MIT and requires the turbo‑tan/llama.cpp‑tq3 fork.
Good at
- Large‑context reasoning
- High‑quality code generation
- Multi‑turn dialogue
Limits
- No multi‑token‑prediction modules
- Requires turbo‑tan/llama.cpp‑tq3 fork
- Experimental build, may have rough edges
Best for
- Chatbots needing 256K context
- Research on large‑context LLMs
- Prototyping with DeepSeek architecture
Who it’s for. Engineers and researchers who need a 150B DeepSeek model with extended context and are comfortable using a custom quantization recipe. The model is suitable for environments that can run the required fork of llama.cpp and handle the larger memory footprint.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.2 GB | 0.8 GB | Fits | Get file → | |
| — | 65.5 GB | 73.9 GB | Too big | Get file → |
The 0.23 GB file uses an unknown quant and needs about 0.8 GB VRAM at 8K context, suitable for 8 GB VRAM setups. The 65.51 GB file is the full TQ3_4S quant and requires roughly 73.9 GB VRAM at 8K context, suitable for high‑end GPUs or CPU with large RAM.
What GPU do you need for DeepSeek V4 Flash 0731 REAP 150b TQ3 4S?
These figures add the file size, an fp16 KV cache at an 8K context window 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 the main file (0.2 GB file, about 0.8 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (0.2 GB file, about 0.8 GB in use).
- 8 GB VRAM: the best quant that fits is the main file (0.2 GB file, about 0.8 GB in use).
- 12 GB VRAM: the best quant that fits is the main file (0.2 GB file, about 0.8 GB in use).
- 16 GB VRAM: the best quant that fits is the main file (0.2 GB file, about 0.8 GB in use).
- 24 GB VRAM: the best quant that fits is the main file (0.2 GB file, about 0.8 GB in use).
- 32 GB VRAM: the best quant that fits is the main file (0.2 GB file, about 0.8 GB in use).
- 48 GB VRAM: the best quant that fits is the main file (0.2 GB file, about 0.8 GB in use).
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about DeepSeek V4 Flash 0731 REAP 150b TQ3 4S
How much VRAM does DeepSeek V4 Flash 0731 REAP 150b TQ3 4S need?
About 0.8 GB at its main file with an 8K context window. That is the 0.2 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 DeepSeek V4 Flash 0731 REAP 150b TQ3 4S on an 8 GB GPU?
Yes, with the smaller files: the 0.2 GB file needs about 0.8 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of DeepSeek V4 Flash 0731 REAP 150b TQ3 4S GGUF?
It depends on the quant: file is 0.2 GB, file is 65.5 GB.
Where can I download DeepSeek V4 Flash 0731 REAP 150b TQ3 4S GGUF?
The files are on Hugging Face in the Tinker-Stack/DeepSeek-V4-Flash-0731-REAP-150b-TQ3_4S-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 DeepSeek V4 Flash 0731 REAP 150b TQ3 4S 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 the 65.51 GB quant on 8 GB VRAM?
No, it requires about 73.9 GB VRAM at 8K context. Use the 0.23 GB file for 8 GB VRAM.
Does this model support vision?
No, the model has no vision capabilities.
Is the quantization uniform?
No, it is a selectively‑quantized TQ3_4S container, not a plain uniform quant.