Back to leaderboard

qwen/qwen3.8-27b

LM-STUDIO MLX

qwen3_5 · 27B · 8bit

Excellent

Aug 18, 2026 · Apple M4 Max

vicuna:7b

OLLAMA GGUF

llama · 7B · Q4_0

Marginal

Mar 1, 2026 · Apple M4

Global Score
81 vs 59
Hardware Fit
68 vs 97
Quality Score
87 vs 43

Hardware

qwen/qwen3.8-27b vicuna:7b
MachineMacBook ProMacBook Air
CPUApple M4 MaxApple M4
Cores1610
RAM128 GB32 GB
GPUApple M4 MaxApple M4
OSmacOS 26.5.2macOS 26.3
Archarm64arm64
Power Modebalancedbalanced

Performance

qwen/qwen3.8-27b vicuna:7b
Tokens/sec16.423.4
First chunk5 msN/A
TTFT742 ms242 ms
Load timeN/A0.4 s
Memory usage38.5 GB5.8 GB
Memory %30%18%

HW Fit Score Breakdown

qwen/qwen3.8-27b

Speed
18/50
TTFT
20/20
Memory
30/30

vicuna:7b

Speed
37/50
TTFT
30/20
Memory
30/30

Quality

qwen/qwen3.8-27b

Reasoning
17/20
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

vicuna:7b

Reasoning
8/20
Coding
4/20
Instruction
10/20
Structured
12/15
Math
2/15
Multilingual
7/10
Reasoning: Weak Coding: Poor Instruction Following: Adequate Structured Output: Strong Math: Poor Multilingual: Adequate

Run yours and compare

$ npm install -g metrillm@latest
$ metrillm

Requires Node 20+ and Ollama or LM Studio running

Or run without installing: npx metrillm@latest