# Phi-4-reasoning: RAM and VRAM requirements

> Phi-4-reasoning is a 14B Phi-4 model. At Q4_K_M it needs about **10.1 GB** to run and fits **29 of 40** tracked devices. Minimum to run: Nvidia GeForce RTX 3060 (12GB).

Last validated: 2026-08-03. Sources: Ollama, HuggingFace GGUF repos, vendor specs.

## Memory by quantization
| Quant | On disk | To run (4k context) |
| --- | --- | --- |
| Q4_K_M | 8.43 GB | ~10.1 GB |
| Q8_0 | 14.51 GB | ~16.2 GB |
| FP16 | 29.32 GB | ~31 GB |

Memory = weights + KV cache + ~0.8 GB runtime overhead, and varies ±15% with context length.

## Will it run on my device?
- **Apple M1 (8GB)** (8 GB): No, not enough memory
- **iPhone 17** (8 GB): No, not enough memory
- **iPhone 17 Pro** (12 GB): No, not enough memory
- **16GB RAM Laptop (CPU/iGPU only)** (16 GB): Yes, it runs
- **Google Pixel 10 Pro** (16 GB): Yes, but tight
- **Nvidia GeForce RTX 3090 (24GB)** (24 GB): Yes, it runs (room for Q8_0)
- **Apple M4 Pro (48GB)** (48 GB): Yes, it runs (room for FP16)
- **Apple M3 Ultra (256GB)** (256 GB): Yes, it runs (room for FP16)

Full table of all 40 devices: https://localmodel.run/model/phi-4-reasoning

## How to run
Quickest path: `ollama run phi4-reasoning:14b`. On Mac, LM Studio (ships MLX) is fastest; on Linux, Ollama for chat or vLLM to serve; on Windows, LM Studio or Ollama.

## Details
- Parameters: 14B
- Default context: 32k tokens
- License: MIT (commercial use: yes)
- Released: 2025-04
- HuggingFace: 23,666 downloads/mo, 228 likes

## FAQ
### How much VRAM or RAM does Phi-4-reasoning need?
About 10.1 GB at Q4_K_M (weights 8.43 GB + KV cache + overhead) at a 4k context. Budget ~16.2 GB for Q8_0.
### Can Phi-4-reasoning run on a laptop?
Phi-4-reasoning is large; you need a high-memory Mac or a 24 GB+ GPU at Q4_K_M.
### Can I use Phi-4-reasoning commercially?
Yes, MIT permits commercial use.

Sources: https://huggingface.co/microsoft/Phi-4-reasoning, https://huggingface.co/bartowski/microsoft_Phi-4-reasoning-GGUF, https://ollama.com/library/phi4-reasoning
More: https://localmodel.run/model/phi-4-reasoning