SShogo Bench

OSWorld 2.0

Real desktop apps across Ubuntu, Windows and macOS, driven from screenshots.

Agent Index weight
6%
in the Computer use pillar
Models tested
9
from the benchmark’s leaderboard and reports
Normalised for the index
0 → 0, 1 → 100
Official site

Results

#ModelScoreNormalisedHarnessCost/taskDate
1Claude Opus 5 Supported
Anthropic
31.4%source——
2GPT-5.6 Sol Supported
OpenAI
27.3%source——
3Claude Opus 4.8 Supported
Anthropic
20.6%source——
4Claude Opus 4.7 Supported
Anthropic
18.2%source——
5GPT-5.5 Supported
OpenAI
13.0%source——
6Claude Sonnet 4.6 Supported
Anthropic
9.3%source——
7Kimi K2.6 Estimated
Moonshot AI
4.6%source——
8MiniMax-M3 Estimated
MiniMax
4.6%source——
9Qwen3.7 Plus
Alibaba (Qwen)
2.8%source——

For models without a OSWorld 2.0 result, the index estimates one from Epoch’s Capabilities Index using a logistic fit on 9 models (residual SD 0.3 on the logit scale). Estimates are labelled and widen the model’s range.