qwen2.5-0.5b-instruct-mlx

qwen2 · 0.5B · 4bit

MacBook Air (Apple M4)

32 GB · macOS 26.3

Tested on March 4, 2026

Top 80% Compare

Global Score

52 /100

Marginal

Hardware Fit

100/100

Quality

31/100

Get this model

Get it in LM Studio

Search and download models directly from the app

🤗

Find on HuggingFace

MLX versions & conversions

Hardware

Machine: MacBook Air
CPU: Apple M4
Cores: 10 total (4 perf + 6 eff)
Frequency: 2.4 GHz
RAM: 32 GB LPDDR5
GPU: Apple M4
OS: macOS 26.3
Arch: arm64
Power Mode: balanced

Performance

Tokens/sec Tokens generated per second — higher is better: 257.5
Standard deviation Variation of speed across benchmark runs — lower means more consistent: ±3.0
First chunk latency Delay before the first streamed chunk arrives from runtime: 99 ms
Time to first token How fast the model starts responding — lower is better: 99 ms
Load time Time to load the model into memory before inference: N/A
Memory usage RAM consumed during inference — percentage of total system memory: 0.3 GB (1%)
Total tokens Total number of tokens generated across all benchmark prompts: 1269

Score breakdown

Speed

50/50

Time to first token

20/20

Memory

30/30

Quality

Reasoning

4/20

Coding

2/20

Instruction following

10/20

Structured output

10/15

Math

2/15

Multilingual

3/10

Category levels

Reasoning: Poor Coding: Poor Instruction Following: Adequate Structured Output: Adequate Math: Poor Multilingual: Weak

Metadata

Spec version: 0.2.1
Runtime: LM Studio 0.4.6+1
Model format: MLX
Hardware profile: BALANCED
Result hash: 1ee47a958dfabe09a066aa38077c6e3d0da35e0341e0a147ce7a1356a55e6408

Interpretation

Hardware fit: 100/100. Overall suitability: MARGINAL (Global 52/100). Category profile: Reasoning: Poor, Coding: Poor, Instruction Following: Adequate, Structured Output: Adequate, Math: Poor, Multilingual: Weak.

Bench Environment

Power: AC CPU load: avg 18% (peak 23%)

Run yours now

$ npm install -g metrillm@latest

$ metrillm

Requires Node 20+ and Ollama or LM Studio running

Or run without installing: npx metrillm@latest