/

← All benchmarks

OSWorld

agentic · computer-use↗ Quelle

63Evidence

not yet measured

What it measures

Ob ein Agent Aufgaben in einer echten Desktop-Umgebung erledigt.

What it does not measure

Textqualitaet, Wissen.

How the score is composed

Currency50

Discrimination50

Headroom50

Contamination safety60

Gaming resistance100

Methodological rigor80

Reproducibility60

Details

Benchmaxxing risk: low
Difficulty: high
Public test set: yes
Human-verified: yes
LLM-judged: no
Skala: 0–100 % solved (höher besser)

Hand-maintained — no automatic adapter yet.

Best measured models

No results ingested yet.