VOOZH about

URL: https://willitrunai.com/models/phi-4-14b

⇱ Phi-4 14B VRAM Requirements β€” GPU Compatibility


πŸ‘ Microsoft
Microsoft

Phi-4 14B

Current
πŸ‘ huggingface
HuggingFaceπŸ‘ ollama
Ollama
891.5KDownloads2.3KLikesDec 2024Released16K tokensContextMITLicense75 StrongQuality

Phi-4 14B (14B parameters) requires approximately 13.4 GB of VRAM with Q4_K_M quantization. For the best balance of quality and speed, we recommend hardware with at least 16 GB of VRAM.

Get started

β€” copy & paste to run locally

Copy-paste commands to run Phi-4 14B on your machine.

Run

ollama run phi4

Quick specs

Parameters14B
Architecturedense
Context16K tokens
Modalitytext
Min RAM5.5 GB
Rec. RAM8.5 GB (Q4_K_M)
LicenseMIT
FamilyPhi
βœ“ Chatβœ“ Reasoning

About this model

Our training data is an extension of the data used for Phi-3 and includes a wide variety of sources from:

  • β€’Publicly available documents filtered rigorously for quality, selected high-quality educational data, and code
  • β€’Newly created synthetic, β€œtextbook-like” data for the purpose of teaching math, coding, common sense reasoning, general knowledge of the world...
  • β€’Acquired academic books and Q&A datasets
  • β€’High quality chat format supervised data covering various topics to reflect human preferences on different aspects such as instruct-following,...

Related models

Your hardware

Detecting...

Quick picks

Best budgetA
RX 7600 XT 16GB~$329 β€” 21 tok/s
πŸ‘ NVIDIA
Best overallS
RTX A4500 20GB~$2,000 β€” 63 tok/s

Best hardware

Top picks for Phi-4 14B

RTX A4500 20GBS
20 GB
RX 7900 XT 20GBS
20 GB
RTX 4090 24GBS
24 GB
NVIDIA A30 24GBS
24 GB
RTX 5090 Laptop 24GBS
24 GB

Run this model

Phi-4 14B on RTX A4500 20GBPhi-4 14B on RX 7900 XT 20GBPhi-4 14B on RTX 4090 24GB

Quantization options

VRAM estimates by quant level

No hardware detected β€” fit column shows raw VRAM estimates

QuantBitsVRAMQualityFit
Q2_K
2
5.5 GB
Lowβ€”
Q3_K_S
3
6.9 GB
Lowβ€”
NVFP4
4
7.8 GB
Mediumβ€”
Q4_K_M
4
8.5 GB
Mediumβ€”
Q5_K_M
5
10.1 GB
Highβ€”
Q6_K
6
11.5 GB
Highβ€”
Q8_0
8
15.0 GB
Very Highβ€”
F16
16
28.7 GB
Maximumβ€”

Quality benchmarks

Phi-4 14B benchmark scores

Benchmark verified

Coding

SWE-bench Verifiedβ€”
HumanEval+82.6%
Aider Polyglotβ€”
LiveCodeBenchβ€”

Reasoning

MMLU-Pro70.4%
GPQA Diamond56.1%
MATH-50080.4%
ARC Challengeβ€”

General

Chatbot Arenaβ€”
IFEval63.0%

Source: official Β· 2024-12-12

Hardware compatibility

Fit estimates across all hardware

Open calculator

Computing compatibility...

Memory breakdown

Reference: RTX 2060 6GB

Weights8.5 GB
KV Cache3.1 GB
Runtime1.2 GB
Headroom0.6 GB

Frequently asked questions

FAQ β€” Phi-4 14B

See also

Quantization GuideScoring MethodologyVRAM Calculator