Back to leaderboard

qwen/qwen3.8-27b

LM-STUDIO MLX

qwen3_5 · 27B · 4bit

Excellent

Aug 19, 2026 · Apple M4 Max

llama3.2:3b

OLLAMA GGUF

llama · 3.2B · Q4_K_M

Good

Jun 25, 2026 · Cortex-A78AE

Global Score
83 vs 74
Hardware Fit
68 vs 99
Quality Score
90 vs 63

Hardware

qwen/qwen3.8-27b llama3.2:3b
MachineMacBook ProNVIDIA NVIDIA Jetson Orin Nano Engineering Reference Developer Kit Super
CPUApple M4 MaxCortex-A78AE
Cores166
RAM128 GB7 GB
GPUApple M4 MaxIntegrated / Unknown
OSmacOS 26.5.2Ubuntu 22.04.5 LTS
Archarm64arm64
Power Modebalancedbalanced

Performance

qwen/qwen3.8-27b llama3.2:3b
Tokens/sec27.621.1
First chunk10 ms1032 ms
TTFT4.6 s1.0 s
Load timeN/A8.4 s
Memory usage21.0 GB2.4 GB
Memory %16%32%

HW Fit Score Breakdown

qwen/qwen3.8-27b

Speed
32/50
TTFT
6/20
Memory
30/30

llama3.2:3b

Speed
50/50
TTFT
20/20
Memory
29/30

Quality

qwen/qwen3.8-27b

Reasoning
19/20
Coding
15/20
Instruction
16/20
Structured
15/15
Math
15/15
Multilingual
10/10
Reasoning: Strong Coding: Adequate Instruction Following: Strong Structured Output: Strong Math: Strong Multilingual: Strong

llama3.2:3b

Reasoning
10/20
Coding
14/20
Instruction
15/20
Structured
12/15
Math
3/15
Multilingual
9/10
Reasoning: Adequate Coding: Adequate Instruction Following: Strong Structured Output: Strong Math: Poor Multilingual: Strong

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