qwen/qwen3.8-27b
LM-STUDIO MLXqwen3_5 · 27B · 4bit
Aug 18, 2026 · Apple M4 Max
qwen3:30b
OLLAMA GGUFqwen3moe · 30.5B · Q4_K_M
Mar 5, 2026 · Intel Core™ Ultra 9 285K
Global Score
87 vs 86
Hardware Fit
83 vs 98
Quality Score
88 vs 81
Hardware
qwen/qwen3.8-27b qwen3:30b
MachineMacBook ProASUS
CPUApple M4 MaxIntel Core™ Ultra 9 285K
Cores1624
RAM128 GB47 GB
GPUApple M4 MaxNVIDIA GeForce RTX 5090, Intel(R) Graphics
OSmacOS 26.5.2Microsoft Windows 11 Pro 10.0.26200
Archarm64x64
Power Modebalancedbalanced
Performance
qwen/qwen3.8-27b qwen3:30b
Tokens/sec29.1204.8
First chunk9 ms96 ms
TTFT650 ms674 ms
Load timeN/A5.2 s
Memory usage21.0 GB20.5 GB
Memory %16%43%
HW Fit Score Breakdown
qwen/qwen3.8-27b
Speed
33/50
TTFT
20/20
Memory
30/30
qwen3:30b
Speed
50/50
TTFT
20/20
Memory
28/30
Quality
qwen/qwen3.8-27b
Reasoning
18/20
Coding
18/20
Instruction
16/20
Structured
15/15
Math
11/15
Multilingual
10/10
Reasoning: Strong Coding: Strong Instruction Following: Strong Structured Output: Strong Math: Adequate Multilingual: Strong
qwen3:30b
Reasoning
16/20
Coding
12/20
Instruction
17/20
Structured
12/15
Math
14/15
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