qwen3.5:4b

qwen35 · 4.2B · Q4_K_M

THINKING MODEL

Microsoft WSL (AMD Ryzen 5 5500)

8 GB · Ubuntu 26.04.1 LTS

Tested on October 12, 2026 · Submitted by john_yuki

Top 89% Compare

Global Score

50 /100

Marginal

Hardware Fit

93/100

Quality

31/100

Get this model

🦙

Ollama

ollama pull qwen3.5:4b

View on Ollama Library

ollama.com/library/qwen3.5

Get it in LM Studio

Search and download models directly from the app

🤗

Find on HuggingFace

GGUF versions & quantizations

Hardware

Machine: Microsoft WSL
CPU: AMD Ryzen 5 5500
Cores: 12 threads (6 cores)
RAM: 8 GB
GPU: Integrated / Unknown
OS: Ubuntu 26.04.1 LTS
Arch: x64

Performance

Tokens/sec Tokens generated per second — higher is better: 92.6
Standard deviation Variation of speed across benchmark runs — lower means more consistent: ±1.1
First chunk latency Delay before the first streamed chunk arrives from runtime: 323 ms
Time to first token How fast the model starts responding — lower is better: 2.8 s
Load time Time to load the model into memory before inference: 9.0 s
Memory usage RAM consumed during inference — percentage of total system memory: 2.9 GB (38%)
Total tokens Total number of tokens generated across all benchmark prompts: 1429
Thinking tokens (est.) Estimated internal reasoning tokens — models with "thinking" generate more tokens: ~778

Score breakdown

Speed

50/50

Time to first token

15/20

Memory

28/30

Quality

Reasoning

18/20

Coding

1/20

Instruction following

0/20

Structured output

0/15

Math

12/15

Multilingual

0/10

Category levels

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

Metadata

Spec version: 0.2.1
Runtime: Ollama 0.34.0
Model format: GGUF
Hardware profile: ENTRY
Result hash: 42ce2939669cf818b93c52f58ae4c57e20deb070789875d4349e7dee99a91486

Interpretation

Hardware fit: 93/100. Overall suitability: MARGINAL (Global 50/100). Category profile: Reasoning: Strong, Coding: Poor, Instruction Following: Poor, Structured Output: Poor, Math: Strong, Multilingual: Poor.

Bench Environment

CPU load: avg 30% (peak 31%)

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