Phi 3 medium 128k instruct needs about 10.7 GB of memory at Q4_K_M (a 8.6 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 10 quants, from 6.1 GB (Q3_K_S) to 14.8 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 6.7 GB.
Phi 3 medium 128k instruct is a 14B-parameter Phi model in GGUF format, quantized by NightFuryTimo. Its trained context length is 128K tokens. The repository has 10 quant files, from 6.1 GB (Q3_K_S) to 14.8 GB (Q8_0).
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q3_K_S | 6.1 GB | 8.2 GB | Too big | Get file → | |
| Q3_K_M | 6.9 GB | 9.1 GB | Too big | Get file → | |
| Q3_K_L | 7.5 GB | 9.7 GB | Too big | Get file → | |
| IQ4_NL | 8.0 GB | 10.1 GB | Too big | Get file → | |
| Q4_K_S | 8.0 GB | 10.1 GB | Too big | Get file → | |
| Q4_K_M | 8.6 GB | 10.7 GB | Too big | Get file → | |
| Q5_K_S | 9.6 GB | 11.8 GB | Too big | Get file → | |
| Q5_K_M | 10.1 GB | 12.3 GB | Too big | Get file → | |
| Q6_K | 11.5 GB | 13.6 GB | Too big | Get file → | |
| Q8_0 | 14.8 GB | 17.0 GB | Too big | Get file → |
What GPU do you need for Phi 3 medium 128k instruct?
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 Q5_K_S (9.6 GB file, about 11.8 GB in use).
- 16 GB VRAM: the best quant that fits is Q6_K (11.5 GB file, about 13.6 GB in use).
- 24 GB VRAM: the best quant that fits is Q8_0 (14.8 GB file, about 17.0 GB in use).
Which Phi 3 medium 128k instruct quant should you choose?
Q4_K_M is the usual starting point for Phi 3 medium 128k instruct: it balances file size and quality well. Here it is a 8.6 GB file.
The Q8_0 file is 14.8 GB, about 1.7 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 Q3_K_S at 6.1 GB, with the quality trade-offs that come with fewer bits. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Frequently asked questions about Phi 3 medium 128k instruct
How much VRAM does Phi 3 medium 128k instruct need?
About 10.7 GB at Q4_K_M with an 8K context window. That is the 8.6 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 Phi 3 medium 128k instruct on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 8.2 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Phi 3 medium 128k instruct GGUF?
It depends on the quant: Q3_K_S is 6.1 GB, Q3_K_M is 6.9 GB, Q3_K_L is 7.5 GB, IQ4_NL is 8.0 GB, Q4_K_S is 8.0 GB, Q4_K_M is 8.6 GB and 4 more.
What context length does Phi 3 medium 128k instruct support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 1.7 GB at 8K and 6.7 GB at 32K.
Where can I download Phi 3 medium 128k instruct GGUF?
The files are on Hugging Face in the NightFuryTimo/Phi-3-medium-128k-instruct-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 Phi 3 medium 128k instruct 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.