mistral:latest
OLLAMA GGUFllama · 7.2B · Q4_K_M
Mar 2, 2026 · Apple M4
qwen3:4b
OLLAMA GGUFqwen3 · 4.0B · Q4_K_M
Jun 26, 2026 · Cortex-A78AE
Global Score
73 vs 32
Hardware Fit
96 vs 98
Quality Score
63 vs 4
Hardware
mistral:latest qwen3:4b
MachineMacBook AirNVIDIA NVIDIA Jetson Orin Nano Engineering Reference Developer Kit Super
CPUApple M4Cortex-A78AE
Cores106
RAM32 GB7 GB
GPUApple M4Integrated / Unknown
OSmacOS 26.3Ubuntu 22.04.5 LTS
Archarm64arm64
Power Modebalancedbalanced
Performance
mistral:latest qwen3:4b
Tokens/sec22.517.9
First chunkN/A950 ms
TTFT257 ms950 ms
Load time0.6 s9.7 s
Memory usage10.1 GB3.0 GB
Memory %32%40%
HW Fit Score Breakdown
mistral:latest
Speed
36/50
TTFT
30/20
Memory
30/30
qwen3:4b
Speed
50/50
TTFT
20/20
Memory
28/30
Quality
mistral:latest
Reasoning
8/20
Coding
12/20
Instruction
15/20
Structured
14/15
Math
5/15
Multilingual
9/10
Reasoning: Weak Coding: Adequate Instruction Following: Strong Structured Output: Strong Math: Weak Multilingual: Strong
qwen3:4b
Reasoning
4/20
Coding
0/20
Instruction
0/20
Structured
0/15
Math
0/15
Multilingual
0/10
Reasoning: Poor Coding: Poor Instruction Following: Poor Structured Output: Poor Math: Poor Multilingual: Poor
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