gemma4:31b

gemma4 · 31.3B · Q4_K_M

THINKING MODEL

MacBook Pro (Apple M4 Max)

64 GB · macOS 26.3.1

Tested on April 13, 2026 · Submitted by sera

Top 34% Compare

Global Score

76 /100

Not Rec.

Hardware Fit

46/100

Quality

89/100

Get this model

🦙

Ollama

ollama pull gemma4:31b

View on Ollama Library

ollama.com/library/gemma4

Get it in LM Studio

Search and download models directly from the app

🤗

Find on HuggingFace

GGUF versions & quantizations

Hardware

Machine: MacBook Pro
CPU: Apple M4 Max
Cores: 16 total (12 perf + 4 eff)
Frequency: 2.4 GHz
RAM: 64 GB LPDDR5
GPU: Apple M4 Max
OS: macOS 26.3.1
Arch: arm64
Power Mode: balanced

Performance

Tokens/sec Tokens generated per second — higher is better: 22.4
Standard deviation Variation of speed across benchmark runs — lower means more consistent: ±0.2
First chunk latency Delay before the first streamed chunk arrives from runtime: 565 ms
Time to first token How fast the model starts responding — lower is better: 30.0 s
Load time Time to load the model into memory before inference: 6.9 s
Memory usage RAM consumed during inference — percentage of total system memory: 43.9 GB (69%)
Total tokens Total number of tokens generated across all benchmark prompts: 1455
Thinking tokens (est.) Estimated internal reasoning tokens — models with "thinking" generate more tokens: ~721

Score breakdown

Speed

30/50

Time to first token

0/20

Memory

16/30

Quality

Reasoning

16/20

Coding

19/20

Instruction following

16/20

Structured output

15/15

Math

14/15

Multilingual

9/10

Category levels

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

Metadata

Spec version: 0.2.1
Runtime: Ollama 0.20.5
Model format: GGUF
Hardware profile: HIGH-END
Result hash: 8471657cc54b3e1608e0c31391077188ec40a5382004742a4b7687e3e2864a19

Interpretation

Hardware fit: 46/100. Overall suitability: NOT RECOMMENDED (Global 76/100). Category profile: Reasoning: Strong, Coding: Strong, Instruction Following: Strong, Structured Output: Strong, Math: Strong, Multilingual: Strong.

Disqualifiers

Time to first token too high: 30000ms (maximum: 15727ms for HIGH-END profile)

Bench Environment

Power: AC CPU load: avg 3% (peak 5%)

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