Back to leaderboard

microsoft/phi-4-mini-reasoning

LM-STUDIO MLX

phi3 · 3.8B · 4bit

Good

Mar 3, 2026 · Apple M4

gemma4:12b

OLLAMA GGUF

gemma4 · 11.9B · Q4_K_M

Not Rec.

Aug 22, 2026 · Apple M1 Pro

Global Score
62 vs 63
Hardware Fit
99 vs 62
Quality Score
46 vs 63

Hardware

microsoft/phi-4-mi… gemma4:12b
MachineMacBook AirMacBook Pro
CPUApple M4Apple M1 Pro
Cores1010
RAM32 GB16 GB
GPUApple M4Apple M1 Pro
OSmacOS 26.3macOS 26.6.2
Archarm64arm64
Power Modebalancedbalanced

Performance

microsoft/phi-4-mi… gemma4:12b
Tokens/sec40.315.3
First chunk289 ms1142 ms
TTFT289 ms30.0 s
Load timeN/A5.4 s
Memory usage2.0 GB7.6 GB
Memory %6%47%

HW Fit Score Breakdown

microsoft/phi-4-mini-re…

Speed
50/50
TTFT
20/20
Memory
29/30

gemma4:12b

Speed
36/50
TTFT
0/20
Memory
26/30

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

gemma4:12b

Reasoning
10/20
Coding
10/20
Instruction
9/20
Structured
14/15
Math
10/15
Multilingual
10/10
Reasoning: Weak Coding: Adequate Instruction Following: Weak Structured Output: Strong Math: Adequate Multilingual: Strong

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