Phi-4-reasoning needs about 11.4 GB of memory at Q4_K_M (a 9.2 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 30 quants, from 3.6 GB (TQ1_0) to 29.3 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 6.7 GB.
Phi-4-reasoning: 14.7B parameter model for coding and reasoning
Phi-4-reasoning is a 14.7B parameter Phi family model built on phi3 architecture. It supports 32768 token context and is released under MIT license. The model excels at code generation and logical reasoning tasks.
Good at
- Coding tasks
- Logical reasoning
- Educational content
Limits
- No vision support
- Requires large VRAM for higher precision
- No multilingual support listed
Best for
- Developers building coding assistants
- Educators creating math problem generators
- Researchers testing reasoning benchmarks
Who it’s for. The model is suited for users who need a large language model focused on code and reasoning. It is ideal for developers and researchers who can allocate sufficient GPU memory and prefer an open‑source, MIT‑licensed solution.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| TQ1_0 | 3.6 GB | 5.8 GB | Fits | Get file → | |
| IQ1_S | 4.0 GB | 6.2 GB | Fits | Get file → | |
| IQ1_M | 4.2 GB | 6.4 GB | Fits | Get file → | |
| TQ2_0 | 4.2 GB | 6.4 GB | Fits | Get file → | |
| IQ2_XXS | 4.5 GB | 6.7 GB | Fits | Get file → | |
| IQ2_XS | 4.9 GB | 7.0 GB | Fits | Get file → | |
| Q2_K_S | 5.3 GB | 7.5 GB | Tight | Get file → | |
| IQ2_S | 5.5 GB | 7.7 GB | Tight | Get file → | |
| IQ2_M | 5.8 GB | 8.0 GB | Tight | Get file → | |
| IQ3_XXS | 6.0 GB | 8.2 GB | Too big | Get file → | |
| IQ3_XS | 6.3 GB | 8.5 GB | Too big | Get file → | |
| IQ3_S | 6.6 GB | 8.7 GB | Too big | Get file → | |
| Q3_K_S | 6.6 GB | 8.7 GB | Too big | Get file → | |
| IQ3_M | 7.0 GB | 9.2 GB | Too big | Get file → | |
| Q3_K_M | 7.6 GB | 9.7 GB | Too big | Get file → | |
| IQ4_XS | 7.9 GB | 10.1 GB | Too big | Get file → | |
| Q4_0 | 8.3 GB | 10.4 GB | Too big | Get file → | |
| IQ4_NL | 8.4 GB | 10.6 GB | Too big | Get file → | |
| Q4_K_S | 8.6 GB | 10.8 GB | Too big | Get file → | |
| Q4_1 | 9.2 GB | 11.3 GB | Too big | Get file → | |
| Q4_K_M | 9.2 GB | 11.4 GB | Too big | Get file → | |
| Q5_0 | 10.1 GB | 12.3 GB | Too big | Get file → | |
| Q5_K_S | 10.2 GB | 12.4 GB | Too big | Get file → | |
| Q5_K_M | 10.7 GB | 12.9 GB | Too big | Get file → | |
| Q5_1 | 11.0 GB | 13.2 GB | Too big | Get file → | |
| Q6_K_M | 12.0 GB | 14.2 GB | Too big | Get file → | |
| Q8_0 | 15.6 GB | 17.8 GB | Too big | Get file → | |
| BF16 | 16.5 GB | 18.7 GB | Too big | Get file → | |
| F16 | 16.5 GB | 18.7 GB | Too big | Get file → | |
| BF16 | 29.3 GB | 31.5 GB | Too big | Get file → |
For 8 GB VRAM, use TQ1_0 (3.59 GB file, 5.8 GB memory at 8 k context) or IQ1_S (3.98 GB file, 6.2 GB memory). For 12 GB VRAM, Q4_K_M (9.19 GB file, 11.4 GB memory) is suitable. Higher precision models such as Q8_0 require 15.58 GB file and 17.8 GB memory, exceeding 12 GB VRAM.
What GPU do you need for Phi-4-reasoning?
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: the best quant that fits is TQ1_0 (3.6 GB file, about 5.8 GB in use).
- 8 GB VRAM: the best quant that fits is IQ2_M (5.8 GB file, about 8.0 GB in use).
- 12 GB VRAM: the best quant that fits is Q4_K_M (9.2 GB file, about 11.4 GB in use).
- 16 GB VRAM: the best quant that fits is Q6_K_M (12.0 GB file, about 14.2 GB in use).
- 24 GB VRAM: the best quant that fits is BF16 (16.5 GB file, about 18.7 GB in use).
- 32 GB VRAM: the best quant that fits is BF16 (29.3 GB file, about 31.5 GB in use).
Which Phi-4-reasoning quant should you choose?
Q4_K_M is the usual starting point for Phi-4-reasoning: it balances file size and quality well. Here it is a 9.2 GB file.
The Q8_0 file is 15.6 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 TQ1_0 at 3.6 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 Phi-4-reasoning
How much VRAM does Phi-4-reasoning need?
About 11.4 GB at Q4_K_M with an 8K context window. That is the 9.2 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 Phi-4-reasoning on an 8 GB GPU?
Yes, with IQ2_M: the 5.8 GB file needs about 8.0 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Phi-4-reasoning GGUF?
It depends on the quant: TQ1_0 is 3.6 GB, IQ1_S is 4.0 GB, IQ1_M is 4.2 GB, TQ2_0 is 4.2 GB, IQ2_XXS is 4.5 GB, IQ2_XS is 4.9 GB and 24 more.
What context length does Phi-4-reasoning support?
The GGUF header lists a trained context length of 32,768 tokens (32K). 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-4-reasoning GGUF?
The files are on Hugging Face in the Mungert/Phi-4-reasoning-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-4-reasoning 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 Phi-4-reasoning on 8 GB VRAM?
Yes, using TQ1_0 or IQ1_S quantizations which require about 5.8–6.2 GB memory at 8 k context.
Does Phi-4-reasoning support vision tasks?
No, the model does not include vision capabilities.