gemma4:26b

gemma4 · 25.2B · Q4_K_M

THINKING MODEL

System manufacturer (AMD Ryzen 9 5950X 16-Core Processor)

32 GB · Microsoft Windows 11 Pro 10.0.26300

Tested on October 2, 2026 · Submitted by Qwen3.8vsGemma4

Top 26% Compare

Global Score

84 /100

Not Rec.

Hardware Fit

80/100

Quality

86/100

Get this model

🦙

Ollama

ollama pull gemma4:26b

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: System manufacturer
CPU: AMD Ryzen 9 5950X 16-Core Processor
Cores: 32 total (32 perf)
Frequency: 3.4 GHz
RAM: 32 GB DDR4
GPU: NVIDIA GeForce RTX 3090
OS: Microsoft Windows 11 Pro 10.0.26300
Arch: x64
Power Mode: balanced

Performance

Tokens/sec Tokens generated per second — higher is better: 147.6
Standard deviation Variation of speed across benchmark runs — lower means more consistent: ±7.6
First chunk latency Delay before the first streamed chunk arrives from runtime: 139 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: 21.3 s
Memory usage RAM consumed during inference — percentage of total system memory: 1.7 GB (5%)
Total tokens Total number of tokens generated across all benchmark prompts: 1460
Thinking tokens (est.) Estimated internal reasoning tokens — models with "thinking" generate more tokens: ~751

Score breakdown

Speed

50/50

Time to first token

0/20

Memory

30/30

Quality

Reasoning

17/20

Coding

14/20

Instruction following

18/20

Structured output

15/15

Math

13/15

Multilingual

9/10

Category levels

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

Metadata

Spec version: 0.2.1
Runtime: Ollama 0.35.0
Model format: GGUF
Hardware profile: HIGH-END
Result hash: ca8f5c9b065bd3ddcdf566742f15fe0cea511de9b54f134d220bac4973362d26

Interpretation

Hardware fit: 80/100. Overall suitability: NOT RECOMMENDED (Global 84/100). Category profile: Reasoning: Strong, Coding: Adequate, Instruction Following: Strong, Structured Output: Strong, Math: Strong, Multilingual: Strong.

Disqualifiers

Time to first token too high: 30000ms (maximum: 15463ms for HIGH-END profile)

Bench Environment

CPU load: avg 55% (peak 58%)

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