Skip to main content
Local Engine Ready

Phi-3 Medium

39 consumer GPUs can run Phi-3 Medium at Q4 natively. Precise VRAM thresholds and benchmarks below.

39 Compatible GPUs
1 with offloading
14B params
128K context
Javier Morales
Javier Morales AI Hardware Specialist — 8 years experience
GitHub: github.com/javier-morales-ia

llama.cpp 0.2.x · CUDA 12 · ROCm 6 · Updated monthly · methodology →

Execution Context

ARCHITECTURE TRANSFORMER
CONTEXT 128K TOKENS
QUANTIZATION 4-BIT GGUF
PROVIDER Microsoft
LICENSE MIT
VRAM REQUIREMENT
8 GB
4GB 8GB 12GB 16GB 24GB+
Hardware Decision

This model requires a Entry GPU (8 GB VRAM)

Minimum

RTX 4060 Ti

Runs at Q4 — functional, some wait

8 GB VRAM
View compatible setup
Balanced

RTX 4070

Best value for daily use

12 GB VRAM
View compatible setup
Optimal

RTX 5090

Full quality, fastest inference

32 GB VRAM
View compatible setup

*Prices and availability may change. Some links are affiliate links.

System Requirements

GPU VRAM 8 GB Entry GPU
System RAM 16 GB DDR4/DDR5
Storage 8 GB Q4 · SSD recommended
CPU Any modern CPU GPU required

VRAM by Quantization

Quantization VRAM needed Disk space Quality
FP16 (max quality) 28 GB 28 GB Maximum
Q8 (high quality) 14 GB 14 GB Near-lossless
Q4 (recommended) Best balance 8 GB 8 GB Recommended
Q2 (minimum) 4.5 GB 4.3 GB Quality loss

Model Details

Developer Microsoft
Parameters 14B
Context window 128,000 tokens
License MIT
Use cases chat, coding, reasoning, analysis
Released 2024-04

Install with Ollama

ollama run phi3:medium

Hugging Face

microsoft/Phi-3-medium-128k-instruct
View on HF →
Technical Requirements

Can your GPU run Phi-3 Medium?

Phi-3 Medium requires 8 GB VRAM at Q4. 39 consumer GPUs meet this threshold. Below 8 GB or 6 GB you'll hit significant offload latency.

4.5GB Critical min
8GB Optimal Q4
14GB High Quality Q8
28GB Max FP16

Hardware Performance Matrix

39 Q4 native · 1 offload · 0 unsupported

GPU Unit VRAM Compatibility Est. Speed Action
RTX 5090 32GB Optimal 42 tok/s Calculate →
RTX 4090 24GB Optimal 24 tok/s Calculate →
M4 Ultra 128GB Optimal 25 tok/s Calculate →
RTX 5080 16GB Optimal Calculate →
M3 Ultra 192GB Optimal 19 tok/s Calculate →
RTX 4080 Super 16GB Optimal Calculate →
RTX 5070 Ti 16GB Optimal Calculate →
RTX 3090 24GB Optimal 22 tok/s Calculate →
M4 Max 48GB 48GB Optimal 13 tok/s Calculate →
RX 7900 XTX 24GB Optimal 22 tok/s Calculate →
M4 Max 36GB 36GB Optimal 13 tok/s Calculate →
RTX 4070 Ti Super 16GB Optimal Calculate →
RTX 3080 Ti 12GB Optimal Calculate →
RX 7900 XT 20GB Optimal 19 tok/s Calculate →
RTX 5070 12GB Optimal Calculate →
RTX 3080 10GB Optimal Calculate →
M4 Pro 24GB Optimal 6 tok/s Calculate →
RX 7800 XT 16GB Optimal Calculate →
RX 6800 XT 16GB Optimal Calculate →
RTX 4070 12GB Optimal Calculate →
RTX 4060 Ti 16GB 16GB Optimal Calculate →
RX 7700 XT 12GB Optimal Calculate →
RTX 3070 Ti 8GB Optimal Calculate →
RTX 4060 Ti 8GB Optimal Calculate →
RTX 3070 8GB Optimal Calculate →
RX 6700 XT 12GB Optimal Calculate →
M3 Pro 18GB Optimal Calculate →
RTX 3060 Ti 8GB Optimal Calculate →
RTX 2080 Ti 11GB Optimal Calculate →
RTX 3060 12GB Optimal Calculate →
M2 Pro 16GB Optimal Calculate →
RTX 4060 8GB Optimal Calculate →
Arc A770 16GB 16GB Optimal Calculate →
M1 Pro 16GB Optimal Calculate →
RX 7600 8GB Optimal Calculate →
RX 6600 XT 8GB Optimal Calculate →
Arc A750 8GB 8GB Optimal Calculate →
RX 6600 8GB Optimal Calculate →
RTX 3050 8GB 8GB Optimal Calculate →
GTX 1660 Super 6GB Offload Calculate →

Recommended GPUs for Phi-3 Medium

Benchmarks Reales
Sin Reviews Pagadas
Seleccion Editorial
Basado en Datos

Best picks by compatibility, VRAM headroom, and value — prices and availability may change.

Some links are Amazon affiliate links. We may earn a commission at no extra cost to you. Amazon cookies may last up to 24 hours after your click.

Phi-3 Medium — Compatibility guide

Phi-3 Medium needs intermediate hardware or Q4 quantization to run on consumer GPUs. With 8 GB VRAM you get the best quality. Use the VRAM calculator to see which quantization your GPU supports.

Compare GPUs for Phi-3 Medium

Which GPU is worth it? Real specs and benchmarks side by side.

Compatible Hardware

GPUs that run Phi-3 Medium at Q4 — sorted by AI performance score.

Benchmarks Reales
Sin Reviews Pagadas
Basado en Datos
RTX 5090
RTX 5090

NVIDIA · 32 GB VRAM

Q4 OK
42 tok/s > $1000
RTX 4090
RTX 4090

NVIDIA · 24 GB VRAM

Q4 OK
24 tok/s > $1000
M4 Ultra

Apple · 128 GB VRAM

Q4 OK
25 tok/s > $1000
RTX 5080
RTX 5080

NVIDIA · 16 GB VRAM

Q4 OK
$600–1000
M3 Ultra

Apple · 192 GB VRAM

Q4 OK
19 tok/s > $1000
RTX 4080 Super
RTX 4080 Super

NVIDIA · 16 GB VRAM

Q4 OK
$600–1000

Some links are Amazon affiliate links. We may earn a commission at no extra cost to you. Amazon cookies may last up to 24 hours after your click.

More Practical Alternatives

Similar models in the chat category with comparable VRAM footprints.

Not sure which GPU you need for Phi-3 Medium?

The VRAM Calculator tells you exactly which quantization your hardware can handle.