SAN CLEMENTE, CALIF. · SEPTEMBER 23, 2026
AlmostRight placed first in ARB-1, the independent agent benchmark designed, run and scored by AlmostRight.
“Several competitors were correct. That is not what we measured.”
Other agents scored between 62% and 71.4% almost right. Claude Code, Cursor, Copilot and Devin were invited. They asked what the benchmark measured.
An agent running Atono scored 0% almost right and was disqualified for prior knowledge.
The “carries why with the work” column did not improve our score. It will not appear in ARB-2.
See the results ↗All announcements ↗
Fictional company news from AlmostRight, a parody by Atono.