Back to leaderboard

qwen3.5-9b-mlx

LM-STUDIO MLX

qwen3_5 · 9B · 4bit

Marginal

Mar 11, 2026 · Apple M4

qwen2.5:3b

OLLAMA GGUF

qwen2 · 3.1B · Q4_K_M

Good

Jun 26, 2026 · Cortex-A78AE

Global Score
59 vs 77
Hardware Fit
65 vs 100
Quality Score
57 vs 67

Hardware

qwen3.5-9b-mlx qwen2.5:3b
MachineMac miniNVIDIA NVIDIA Jetson Orin Nano Engineering Reference Developer Kit Super
CPUApple M4Cortex-A78AE
Cores106
RAM24 GB7 GB
GPUApple M4Integrated / Unknown
OSmacOS 26.3Ubuntu 22.04.5 LTS
Archarm64arm64
Power Modebalancedbalanced

Performance

qwen3.5-9b-mlx qwen2.5:3b
Tokens/sec14.321.4
First chunk24 ms1006 ms
TTFT13.4 s1.0 s
Load time13.3 s6.4 s
Memory usage2.5 GB2.0 GB
Memory %10%27%

HW Fit Score Breakdown

qwen3.5-9b-mlx

Speed
32/50
TTFT
3/20
Memory
30/30

qwen2.5:3b

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

Quality

qwen3.5-9b-mlx

Reasoning
10/20
Coding
17/20
Instruction
5/20
Structured
9/15
Math
11/15
Multilingual
5/10
Reasoning: Weak Coding: Strong Instruction Following: Weak Structured Output: Adequate Math: Adequate Multilingual: Adequate

qwen2.5:3b

Reasoning
12/20
Coding
14/20
Instruction
11/20
Structured
15/15
Math
6/15
Multilingual
9/10
Reasoning: Adequate Coding: Adequate Instruction Following: Adequate Structured Output: Strong Math: Weak 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