qwen/qwen3.8-27b
LM-STUDIO MLXqwen3_5 · 27B · 4bit
Aug 18, 2026 · Apple M4 Max
openai/gpt-oss-20b
LM-STUDIO GGUFgpt-oss · 20B · MXFP4
Apr 24, 2026 · AMD Ryzen 5 5500
Global Score
87 vs 90
Hardware Fit
83 vs 88
Quality Score
88 vs 91
Hardware
qwen/qwen3.8-27b openai/gpt-oss-20b
MachineMacBook ProASUS
CPUApple M4 MaxAMD Ryzen 5 5500
Cores1612
RAM128 GB31 GB
GPUApple M4 MaxNavi 44 [Radeon RX 9060 XT]
OSmacOS 26.5.2Nobara Linux 43
Archarm64x64
Power Modebalancedperformance
Performance
qwen/qwen3.8-27b openai/gpt-oss-20b
Tokens/sec29.158.7
First chunk9 ms6 ms
TTFT650 ms4.5 s
Load timeN/AN/A
Memory usage21.0 GB11.9 GB
Memory %16%38%
HW Fit Score Breakdown
qwen/qwen3.8-27b
Speed
33/50
TTFT
20/20
Memory
30/30
openai/gpt-oss-20b
Speed
50/50
TTFT
10/20
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
openai/gpt-oss-20b
Reasoning
18/20
Coding
15/20
Instruction
18/20
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