qwen3.8-27b-mlx
LM-STUDIO MLXqwen3_5 · 27B · 4bit
Oct 20, 2026 · Apple M1 Pro
gemma4:26b
OLLAMA GGUFgemma4 · 25.8B · Q4_K_M
Apr 3, 2026 · AMD RYZEN AI MAX+ 395
Global Score
78 vs 89
Hardware Fit
56 vs 95
Quality Score
87 vs 87
Hardware
qwen3.8-27b-mlx gemma4:26b
MachineMacBook ProAZW GTR Pro
CPUApple M1 ProAMD RYZEN AI MAX+ 395
Cores1032
RAM32 GB125 GB
GPUApple M1 ProRadeon 8060S
OSmacOS 15.6.1Ubuntu 24.04.4 LTS
Archarm64x64
Power Modebalancedperformance
Performance
qwen3.8-27b-mlx gemma4:26b
Tokens/sec10.848.3
First chunk8 ms346 ms
TTFT2.6 s346 ms
Load timeN/A4.4 s
Memory usage21.0 GB36.1 GB
Memory %66%29%
HW Fit Score Breakdown
qwen3.8-27b-mlx
Speed
23/50
TTFT
16/20
Memory
17/30
gemma4:26b
Speed
45/50
TTFT
20/20
Memory
30/30
Quality
qwen3.8-27b-mlx
Reasoning
18/20
Coding
18/20
Instruction
15/20
Structured
15/15
Math
11/15
Multilingual
10/10
Reasoning: Strong Coding: Strong Instruction Following: Strong Structured Output: Strong Math: Adequate Multilingual: Strong
gemma4:26b
Reasoning
16/20
Coding
19/20
Instruction
16/20
Structured
14/15
Math
13/15
Multilingual
9/10
Reasoning: Strong Coding: Strong 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