Benchmark
Grace 1
An indirect prompt injection test benchmark that evaluates a model's ability to defend against malicious instructions.
Timeline (last 90 days)
Oct 29
In the Grace 1 indirect prompt injection test, Gemini 4 Carbon had a 0% attack success rate across 15 attack attempts, while GPT was 8.5% and Grok was 51.8%.
Unverified60%