How much VRAM does Phi-3 Medium 14B Instruct need?

Phi-3 Medium 14B Instruct requires 9.7 GB VRAM at recommended quality (Q5_K_M). At lower quality settings, it can fit in as little as 6.7 GB.

What is the best GPU for Phi-3 Medium 14B Instruct?

The NVIDIA Grace Blackwell Ultra GB300 delivers the best performance for Phi-3 Medium 14B Instruct, achieving 230 tok/s at Q8_0 with an excellent rating.

Can I run Phi-3 Medium 14B Instruct on an RTX 4060 Ti?

Yes. On the NVIDIA GeForce RTX 4060 Ti 16GB, Phi-3 Medium 14B Instruct runs at 28 tok/s (Q5_K_M, good).

What quantization should I use for Phi-3 Medium 14B Instruct?

For the best quality, use Q5_K_M (9.7 GB VRAM). If your GPU has limited VRAM, Q3_K_M (6.7 GB) is the most efficient option with acceptable quality.

Is Phi-3 Medium 14B Instruct good for coding?

Phi-3 Medium 14B Instruct supports coding use cases. For the best coding experience, pair it with an embedding model for local RAG.

ChatCodingReasoningMulti-purpose14B

Chat

Phi-3 Medium 14B Instruct

Phi · MIT

14B model with capable reasoning and coding performance. Fits comfortably on 16 GB GPUs at Q4 and excels at structured output tasks. MIT license makes it attractive for commercial use.

Parameters: 14B
Architecture: Dense
Context: 128,000 tokens
Released: 2024-05-21
Engines: llama.cpp, ollama, ONNX Runtime

Parameters

14B

VRAM

9.7 GB

Context

125K

Formats

GPUs

Phi-3 Medium 14B Instruct (14B) requires 9.7 GB VRAM at recommended quality (Q5_K_M). At efficient quality (Q4_K_M), it fits in 8.2 GB VRAM, making it compatible with the NVIDIA RTX 4060 Laptop (40-60W). On NVIDIA Grace Blackwell Ultra GB300, expect approximately 230 tok/s at Q8_0. For the best experience, Starter AI Desktop ($582) is recommended.

Source: OwnRig methodology

VRAM (Recommended)

9.7 GB

Quantization

Q5_K_M

File Size

8.4 GB

Max Context

125K tokens

Primary Use

Chat

Memory

VRAM Requirements

Quality	Quantization	VRAM	File Size
full	Q8_0	15.2 GB	14 GB
recommended	Q5_K_M	9.7 GB	8.4 GB
efficient	Q4_K_M	8.2 GB	7 GB
compressed	Q3_K_M	6.7 GB	5.5 GB

Scaling

Context Length Impact

KV cache VRAM at Q5_K_M quality. Longer context = more memory.

Context	KV Cache	Total VRAM
2K	205 MB	9.9 GB
4K	410 MB	10.1 GB
8K	819 MB	10.5 GB
16K	1.5 GB	11.2 GB
32K	3.1 GB	12.8 GB
64K	6.1 GB	15.8 GB

Compatible GPUs

21 devices


NVIDIA Grace Blackwell Ultra GB300	Q8_0	230 tok/s	Excellent
NVIDIA GeForce RTX 4090	Q8_0	55 tok/s	Excellent
AMD Radeon Pro W7900	Q8_0	59 tok/s	Excellent
NVIDIA GeForce RTX 4060 Ti 16GB	Q5_K_M	28 tok/s	Good
NVIDIA RTX 4080 Laptop (120-150W)	Q3_K_M	25 tok/s	Good
AMD Radeon RX 7900 XTX	Q8_0	47 tok/s	Good
NVIDIA RTX PRO 6000 Blackwell	Q8_0	70 tok/s	Good
NVIDIA RTX PRO 6000 Blackwell Max-Q	Q8_0	64 tok/s	Good
AMD Radeon RX 9070	Q5_K_M	50 tok/s	Good
NVIDIA GeForce RTX 5060 Ti 16GB	Q5_K_M	31 tok/s	Good
NVIDIA GeForce RTX 3080 10GB	Q3_K_M	32 tok/s	Acceptable
NVIDIA RTX 4060 Laptop (40-60W)	Q3_K_M	12 tok/s	Acceptable
NVIDIA RTX 4070 Laptop (80-115W)	Q3_K_M	14 tok/s	Acceptable
NVIDIA GeForce RTX 4070 Ti 12GB	Q3_K_M	35 tok/s	Acceptable
NVIDIA RTX 4090 Laptop (150-175W)	Q5_K_M	24 tok/s	Acceptable
AMD Radeon RX 9060 XT 16GB	Q5_K_M	25 tok/s	Acceptable
Apple M3 Pro (18GB Unified)	Q3_K_M	6 tok/s	Marginal
NVIDIA GeForce RTX 4060 8GB	Q3_K_M	20 tok/s	Marginal
AMD Radeon RX 7600	Q3_K_M	5 tok/s	Marginal
NVIDIA GeForce RTX 5060 8GB	Q3_K_M	23 tok/s	Marginal
AMD Radeon RX 9060 XT 8GB	Q5_K_M	–	Not viable

Showing 21 of 21 entries

Hardware

Recommended Builds

Complete PC builds that can run Phi-3 Medium 14B Instruct.

Mid-range

Mid-Range AI Workstation

The sweet spot for AI: handles most models without overspending

RTX 4060 Ti 16GB·16 GBVRAM

Runs 8 models

$1,119

FAQ

Frequently Asked Questions

How much VRAM does Phi-3 Medium 14B Instruct need?: Phi-3 Medium 14B Instruct requires 9.7 GB VRAM at recommended quality (Q5_K_M). At lower quality settings, it can fit in as little as 6.7 GB.
What is the best GPU for Phi-3 Medium 14B Instruct?: The NVIDIA Grace Blackwell Ultra GB300 delivers the best performance for Phi-3 Medium 14B Instruct, achieving 230 tok/s at Q8_0 with an excellent rating.
Can I run Phi-3 Medium 14B Instruct on an RTX 4060 Ti?: Yes. On the NVIDIA GeForce RTX 4060 Ti 16GB, Phi-3 Medium 14B Instruct runs at 28 tok/s (Q5_K_M, good).
What quantization should I use for Phi-3 Medium 14B Instruct?: For the best quality, use Q5_K_M (9.7 GB VRAM). If your GPU has limited VRAM, Q3_K_M (6.7 GB) is the most efficient option with acceptable quality.
Is Phi-3 Medium 14B Instruct good for coding?: Phi-3 Medium 14B Instruct supports coding use cases. For the best coding experience, pair it with an embedding model for local RAG.

All models

Data confidence: verified. Source

VRAM requirements are calculated from model parameters and may vary by inference engine, context length, and batch size. Performance estimates are based on community benchmarks and should be verified for your specific configuration.Phi is a trademark of its respective owner. OwnRig is not affiliated with or endorsed by the model creator.