microsoft/phi-4-mini-reasoning

LM-STUDIO MLX

phi3 · 3.8B · 4bit

Good

Mar 3, 2026 · Apple M4

deepseek-r1-distill-qwen-14b-mlx

LM-STUDIO MLX

qwen2 · 14B · 5bit

Not Rec.

Mar 11, 2026 · Apple M4

Global Score

62 vs 36

Hardware Fit

99 vs 53

Quality Score

46 vs 28

Hardware

microsoft/phi-4-mi… deepseek-r1-distil…

MachineMacBook AirMac mini

CPUApple M4Apple M4

Cores1010

RAM32 GB24 GB

GPUApple M4Apple M4

OSmacOS 26.3macOS 26.3

Archarm64arm64

Power Modebalancedbalanced

Performance

microsoft/phi-4-mi… deepseek-r1-distil…

Tokens/sec40.310.3

First chunk289 ms27 ms

TTFT289 ms20.7 s

Load timeN/A15.4 s

Memory usage2.0 GB1.0 GB

Memory %6%4%

HW Fit Score Breakdown

microsoft/phi-4-mini-re…

Speed

50/50

TTFT

20/20

Memory

29/30

deepseek-r1-distill-qwe…

Speed

23/50

TTFT

0/20

Memory

30/30

Quality

microsoft/phi-4-mini-re…

Reasoning

7/20

Coding

6/20

Instruction

7/20

Structured

7/15

Math

12/15

Multilingual

7/10

Reasoning: Weak Coding: Weak Instruction Following: Weak Structured Output: Weak Math: Strong Multilingual: Adequate

deepseek-r1-distill-qwe…

Reasoning

3/20

Coding

6/20

Instruction

6/20

Structured

3/15

Math

2/15

Multilingual

8/10

Reasoning: Poor Coding: Weak Instruction Following: Weak Structured Output: Poor Math: Poor Multilingual: Strong

View microsoft/phi-4-mi… details | View deepseek-r1-distil… details

Run yours and compare

$ npm install -g metrillm@latest

$ metrillm

Requires Node 20+ and Ollama or LM Studio running

Or run without installing: npx metrillm@latest