Is OSWorld-Verified still meaningful?
Status:Near ceiling Checked against the report, 5 November 2026
Verdict
No longer for the frontier. Agents have reached the human baseline.
Why near ceiling: Claude Sonnet 4.6 scored 72.1% in February 2026, against a human baseline of 72.36%.[1][1] Benchmark data bundle Epoch AI Primary
What “Near ceiling” meansKnown problems
What to look at instead
- OSWorld 2.0 Status:Still useful Yes. Full-credit scores stay low, though partial credit tells a kinder story.
- Remote Labor Index Status:Still useful Yes. It is one of the few tests where the best agent still fails most tasks.
Where it appears in the report
Chapter 1, Benchmarks
- The new frontier prices tasks in money and hours
- Table 2. First model to cross each threshold in “Go deeper: The full saturation timeline (14 benchmarks)”
- Table 3. How each area replaced its saturated benchmarks, oldest first in “Go deeper: Lineage of replacements, by area”
- Table 7. General and computer-use agent benchmarks, November 2026 in “Go deeper: General and computer-use agent benchmarks (18)”
Chapter 3, Tools and technology
- Eval environments and RL training environments merged into one layer
- Table 4. GitHub API and pypistats, 5 Nov 2026 in “Go deeper: All nine environment projects, plus the benchmark worlds”
Sources
Copied from the chapter source lists. Each number links back to the source in its chapter.
- [1] Benchmark data bundle, Epoch AI. Dated scores (5 Nov 2026) and ceilings for most benchmarks here. Primary epoch.ai Chapter 1, source 1
- [2] What OSWorld tells us, Epoch AI. It analyses how much of OSWorld needs a GUI. Primary epoch.ai Chapter 1, source 53
- [3] OSWorld README, XLANG Lab. It describes OSWorld-Verified and the VM options. Primary github.com Chapter 3, source 115