lmstudio-community/Phi-4-reasoning-plus-MLX-4bit
LM-STUDIO GGUFMar 5, 2026 · Apple M4
qwen/qwen3.8-27b
LM-STUDIO MLXqwen3_5 · 27B · 4bit
Aug 19, 2026 · Apple M4 Max
Global Score
33 vs 83
Hardware Fit
74 vs 68
Quality Score
15 vs 90
Hardware
lmstudio-community… qwen/qwen3.8-27b
MachineMacBook AirMacBook Pro
CPUApple M4Apple M4 Max
Cores1016
RAM32 GB128 GB
GPUApple M4Apple M4 Max
OSmacOS 26.3macOS 26.5.2
Archarm64arm64
Power Modebalancedbalanced
Performance
lmstudio-community… qwen/qwen3.8-27b
Tokens/sec11.527.6
First chunk670 ms10 ms
TTFT670 ms4.6 s
Load timeN/AN/A
Memory usage3.0 GB21.0 GB
Memory %9%16%
HW Fit Score Breakdown
lmstudio-community/Phi-…
Speed
24/50
TTFT
20/20
Memory
30/30
qwen/qwen3.8-27b
Speed
32/50
TTFT
6/20
Memory
30/30
Quality
lmstudio-community/Phi-…
Reasoning
8/20
Coding
1/20
Instruction
1/20
Structured
1/15
Math
1/15
Multilingual
3/10
Reasoning: Weak Coding: Poor Instruction Following: Poor Structured Output: Poor Math: Poor Multilingual: Weak
qwen/qwen3.8-27b
Reasoning
19/20
Coding
15/20
Instruction
16/20
Structured
15/15
Math
15/15
Multilingual
10/10
Reasoning: Strong Coding: Adequate Instruction Following: Strong Structured Output: Strong Math: Strong Multilingual: Strong
Run yours and compare
$
npm install -g metrillm@latest$
metrillmRequires Node 20+ and Ollama or LM Studio running
Or run without installing: npx metrillm@latest