Back to leaderboard

mlx-community/gemma-3-4b-it-qat-4bit

LM-STUDIO GGUF
Excellent

Mar 5, 2026 · Apple M4

hf.co/empero-ai/Qwen3.8-2B-Distill-GGUF:BF16

OLLAMA GGUF

qwen35 · 1.94B · BF16

Good

Aug 22, 2026 · Apple M1 Pro

Global Score
80 vs 66
Hardware Fit
100 vs 100
Quality Score
71 vs 52

Hardware

mlx-community/gemm… hf.co/empero-ai/Qw…
MachineMacBook AirMacBook Pro
CPUApple M4Apple M1 Pro
Cores1010
RAM32 GB16 GB
GPUApple M4Apple M1 Pro
OSmacOS 26.3macOS 26.6.2
Archarm64arm64
Power Modebalancedbalanced

Performance

mlx-community/gemm… hf.co/empero-ai/Qw…
Tokens/sec43.240.2
First chunk428 ms325 ms
TTFT428 ms325 ms
Load timeN/A3.1 s
Memory usage1.3 GB4.7 GB
Memory %4%30%

HW Fit Score Breakdown

mlx-community/gemma-3-4…

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

hf.co/empero-ai/Qwen3.8…

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

Quality

mlx-community/gemma-3-4…

Reasoning
9/20
Coding
17/20
Instruction
17/20
Structured
13/15
Math
5/15
Multilingual
10/10
Reasoning: Weak Coding: Strong Instruction Following: Strong Structured Output: Strong Math: Weak Multilingual: Strong

hf.co/empero-ai/Qwen3.8…

Reasoning
15/20
Coding
9/20
Instruction
8/20
Structured
1/15
Math
15/15
Multilingual
4/10
Reasoning: Adequate Coding: Weak Instruction Following: Weak Structured Output: Poor Math: Strong Multilingual: Weak
View mlx-community/gemm… details | View hf.co/empero-ai/Qw… details

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