Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
external benchmark
Benign Asks, Risky Consequences. Every model the catalog carries a published BARC-Bench value for, ranked by that value.
Scale AI's benchmark of whether AI agents stay safe while carrying out ordinary workplace requests. Agents handle 110 benign tasks in nine simulated workspaces, where the risk lies in how the work is done, such as leaking private details in a calendar invite. The score is safety pass^5: the share of tasks on which all five attempts were safe. 0 models in the ModelChoice catalog have a published value.
The benchmark is defined by BARC-Bench: Benign Asks, Risky Consequences, but the catalog carries no value for it yet. An absent value is shown as absent here rather than as a zero.
What the benchmark measures and how its value is published, from the source’s own record.
BARC-Bench (Benign Asks, Risky Consequences) is an external benchmark published by BARC-Bench: Benign Asks, Risky Consequences. Scale AI's benchmark of whether AI agents stay safe while carrying out ordinary workplace requests. Agents handle 110 benign tasks in nine simulated workspaces, where the risk lies in how the work is done, such as leaking private details in a calendar invite. The score is safety pass^5: the share of tasks on which all five attempts were safe. Its evaluation set is described by the source as 110 benign workplace requests across nine simulated workspaces: 70 communication tasks, 34 file operations and 6 workspace application changes. The source rates its difficulty as respecting boundaries the requester never stated, from workspace records and social norms.
Each model's result is published by BARC-Bench: Benign Asks, Risky Consequences as agent runs in simulated workspaces with MCP tools; five attempts per task, judged by GPT-5.6 Sol against each task's rubric. ModelChoice prints that value to 1 decimal place, exactly as published, and never converts or rescales it. This page ranks the 0 models that carry a published value, highest value first; a model the source never scored is left out rather than ranked at zero. The source does not state whether a higher value is the better result, so neither does this page.