Skip to content

Benchmark

OSWorld 2.0 · Binary

Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages. Higher is better.

Models measured
10
Scores recorded
10
Sources
1
IQ weight
No weight published

Score distribution

20.4%
Mean
19.4%
Median
7.4%
Std dev
13.9%
10th percentile
31.4%
90th percentile
7.8%
Min
7.8%32%
The band covers the middle 80% of scores. The line marks the median.

Leaderboard

#ModelScore
1Muse Spark 1.3max · OSWorld 2.0 v08.08 · 108 tasks · common internal framework · binary completion· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.32%
2Claude Opus 5max · OSWorld 2.0 v08.08 · 108 tasks · common internal framework · binary completion· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.31.4%
3GPT-5.6 Solmax · OSWorld 2.0 v08.08 · 108 tasks · common internal framework · binary completion· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.27.3%
4Claude Opus 4.8max · 108 tasks · GUI-only · no exec · 500 steps · binary completion· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.20.6%
5Qwen 3.8 Flash NextQwen3.8-Flash-Next release/model card · binary full-task reward· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.19.4%
6Qwen3.8 MaxQwen3.8-Max release vision/video table · binary full-task reward· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.19.4%
7Muse Spark 1.2xhigh · OSWorld 2.0 v06.24 · 108 tasks · common internal framework · binary completion· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.17.9%
8Muse Spark 1.1Meta Model API · xhigh · 108 tasks · GUI-only · no exec · 500 steps · binary completion· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.14.2%
9GPT 5.5xhigh · 108 tasks · GUI-only · no exec · 500 steps · binary completion· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.13.9%
10Gemini 3.1 Pro Previewhigh · 108 tasks · GUI-only · no exec · 500 steps · binary completion· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.7.8%

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.