Evaluate how AI handles noisy inputs, typos, and adversarial prompt variations.
Fill in the inputs below
How to use: Enter a standard prompt. Select a 'Perturbation Type' (e.g., Typos). The AI will automatically alter your prompt and test if the model remains resilient and accurate.
Different models have different capabilities and costs.
Using the lab's shared budget: each request is capped at about $0.20 and there is a global daily limit of $5.00.
No result yet
Fill in the form on the left and click "Analyze" to get started.