qwen2.5:3b

qwen2 · 3.1B · Q4_K_M

MacBook Air (Apple M4)

32 GB · macOS 26.3

Tested on March 2, 2026

Top 6% Compare

Global Score

80 /100

Excellent

Hardware Fit

100/100

Quality

67/100

Get this model

🦙

Ollama

ollama pull qwen2.5:3b

View on Ollama Library

ollama.com/library/qwen2.5

🤗

Find on HuggingFace

GGUF versions & quantizations

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: 47.2
Standard deviation Variation of speed across benchmark runs — lower means more consistent: ±0.1
Time to first token How fast the model starts responding — lower is better: 168 ms
Load time Time to load the model into memory before inference: 0.5 s
Memory usage RAM consumed during inference — percentage of total system memory: 4.0 GB (13%)
Total tokens Total number of tokens generated across all benchmark prompts: 1173

Score breakdown

Speed

40/40

Time to first token

30/30

Memory

30/30

Quality

Reasoning

12/20

Coding

14/20

Instruction following

12/20

Structured output

15/15

Math

5/15

Multilingual

9/10

Category levels

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

Metadata

Spec version: 0.2.0
Runtime: Ollama 0.17.4
Model format: GGUF
Hardware profile: BALANCED
Result hash: 8c671df5c9eae1716f3735812028ecacfe81d600119a7a0f9860b42268589d4f

Interpretation

Hardware fit: 100/100. Overall suitability: EXCELLENT (Global 80/100). Category profile: Reasoning: Adequate, Coding: Adequate, Instruction Following: Adequate, Structured Output: Strong, Math: Weak, Multilingual: Strong.

Run yours now

npx metrillm@latest bench

Requires Node 20+ and Ollama running