Card-wide default max effort; default settings at 1080p and up to 500 action steps per task; first-attempt success averaged over five runs; Opus 4.8 grader where required.
未知项
未另记
工具
工具条件未记录
推理预算
max
框架
未记录
协议
OSWorld 2.0; first-attempt success rate.
条件
Average of five runs; default settings, 1080p, up to 500 action steps per task; Opus 4.8 grader where required. Table 8.1.A rounds the score to 70.6%; later Sep 2026 OSWorld update remains separate.
Claude Opus 5 · OSWorld · 2.0;Opus 5 发布表,未注明评测子集/协议 · Scores and configurations
Claude Opus 5 record for OSWorld · 2.0;Opus 5 发布表,未注明评测子集/协议. 1 score record for this release, including 1 exact value (may include historical records). Includes 2 source locations.
Card-wide default max effort; default settings at 1080p and up to 500 action steps per task; first-attempt success averaged over five runs; Opus 4.8 grader where required.
Unknowns
none recorded
Tools
Tool access not recorded
Reasoning budget
max
Framework
not recorded
Protocol
OSWorld 2.0; first-attempt success rate.
Conditions
Average of five runs; default settings, 1080p, up to 500 action steps per task; Opus 4.8 grader where required. Table 8.1.A rounds the score to 70.6%; later Sep 2026 OSWorld update remains separate.