qwen3.5:9b

qwen35 · 9.7B · Q4_K_M

MacBook Air (Apple M4)

32 GB · macOS 26.3

Tested on March 2, 2026

Top 5% Compare

Global Score

80 /100

Excellent

Hardware Fit

81/100

Quality

80/100

Get this model

🦙

Ollama

ollama pull qwen3.5:9b

View on Ollama Library

ollama.com/library/qwen3.5

🤗

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: 11.8
Standard deviation Variation of speed across benchmark runs — lower means more consistent: ±0.5
Time to first token How fast the model starts responding — lower is better: 523 ms
Load time Time to load the model into memory before inference: 2.9 s
Memory usage RAM consumed during inference — percentage of total system memory: 9.1 GB (28%)
Total tokens Total number of tokens generated across all benchmark prompts: 1017

Score breakdown

Speed

21/40

Time to first token

30/30

Memory

30/30

Quality

Reasoning

14/20

Coding

16/20

Instruction following

15/20

Structured output

15/15

Math

10/15

Multilingual

10/10

Category levels

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

Metadata

Spec version: 0.2.0
Runtime: Ollama 0.17.5
Model format: GGUF
Hardware profile: BALANCED
Result hash: e4d07e22c45ed916ebd546a4e2aad543aaed42764cf136be8ca0e4f79401e3f3

Interpretation

Hardware fit: 81/100. Overall suitability: EXCELLENT (Global 80/100). Category profile: Reasoning: Adequate, Coding: Strong, Instruction Following: Strong, Structured Output: Strong, Math: Adequate, Multilingual: Strong.

Run yours now

npx metrillm@latest bench

Requires Node 20+ and Ollama or LM Studio running