Measured model

Measured on a Mac Studio (M5 Max, 36GB)gpt-oss:20b

OpenAI

On a Mac Studio (M5 Max, 36GB), gpt-oss:20b answers a simple question in about 1.8 s, gets 83% of the Japanese knowledge questions right, 98% of the math problems, and uses 12.8GB of memory.

Results

MeasureResultAmong measured modelsNote
Reply timeabout 1.8 s10 of 33Until one simple common-sense question is answered (median of 50)
Knowledge (high school to university)83%8 of 33200 questions, ±6
Math (word problems)98%3 of 33100 questions, ±5
Memory used12.8GB19 of 33
EN→JA translation33 pts7 of 30out of 100 (chrF), about 4 s a sentence
JA→EN translation55 pts14 of 30out of 100 (chrF), about 3.8 s a sentence

"Among measured models" is the place among the 33 models measured on this site (lower is better for time and memory).

Measured on other Macs

The numbers and places on this page come from the Mac above alone.

Try it on your Mac

  1. Install the Mac version of Ollama from ollama.com and start it.
  2. Type this in Terminal. The first time, the model downloads; then you can chat with it.ollama run gpt-oss:20b

It used 12.8GB here. To use it comfortably alongside other apps, a Mac with 32GB of memory or more is a good guide (the model then takes about half the memory).

About the model

Maker
OpenAI
Size
20.9B parameters
Quantization
MXFP4
Thinks before answering
Yes
Longest input at once
131,072 tokens
Released (Hugging Face)
Aug 4, 2025
Generation speed
89.9 tok/s
Start-up (loading)
about 3.6 s
Measured on
Nov 2, 2026
Ollama
0.34.3
Hugging Face
openai/gpt-oss-20b
Series
gpt-oss

What the parts of a name mean (size, quantization and so on): how to read a model name.

Models of similar memory

Compare the main models on the text page. The series page lists every model measured.

Task recommendations and capabilities · Measurements by model