qwen3.8-27b-mlx
LM-STUDIO MLXqwen3_5 · 27B · 4bit
Oct 20, 2026 · Apple M1 Pro
qwen3.6:35b-a3b-q8_0
OLLAMA GGUFqwen35moe · 36.0B · Q8_0
Jul 7, 2026 · Cortex-X925
Global Score
78 vs 80
Hardware Fit
56 vs 79
Quality Score
87 vs 81
Hardware
qwen3.8-27b-mlx qwen3.6:35b-a3b-q8_0
MachineMacBook ProASUSTeK COMPUTER INC. GX10
CPUApple M1 ProCortex-X925
Cores1020
RAM32 GB122 GB
GPUApple M1 ProDevice 2e12
OSmacOS 15.6.1Ubuntu 24.04.4 LTS
Archarm64arm64
Power Modebalancedperformance
Performance
qwen3.8-27b-mlx qwen3.6:35b-a3b-q8_0
Tokens/sec10.855.5
First chunk8 ms528 ms
TTFT2.6 s30.0 s
Load timeN/A32.6 s
Memory usage21.0 GB41.0 GB
Memory %66%34%
HW Fit Score Breakdown
qwen3.8-27b-mlx
Speed
23/50
TTFT
16/20
Memory
17/30
qwen3.6:35b-a3b-q8_0
Speed
50/50
TTFT
0/20
Memory
29/30
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
qwen3.6:35b-a3b-q8_0
Reasoning
18/20
Coding
13/20
Instruction
17/20
Structured
10/15
Math
13/15
Multilingual
10/10
Reasoning: Strong Coding: Adequate Instruction Following: Strong Structured Output: Adequate 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