Bench-Corp
A test suite used by the producers of OpenAPPA to evaluate guardrails. It counts any violating action as an attack and measures task completion and cost overhead.
Timeline (last 90 days)
In Bench-Corp tests, the guardrail missed twice out of 520 runs, both on the same task with Gemini 3.7 Flash.
On Bench-Corp, because the guardrail intervened on nearly every run, the agent's token usage increased 2.6-fold to 8-fold, DeepSeek's run cost was 16 times higher, and the time taken increased 22-fold.
All Facts (2)
In Bench-Corp tests, the guardrail missed twice out of 520 runs, both on the same task with Gemini 3.7 Flash.
50%UnverifiedOn Bench-Corp, because the guardrail intervened on nearly every run, the agent's token usage increased 2.6-fold to 8-fold, DeepSeek's run cost was 16 times higher, and the time taken increased 22-fold.
50%