Pricing
Transparent pricing for every stage of AI security.
A fast security check covering the most common AI vulnerabilities.
- ✓Tier 1 evaluation (5 attack vectors)
- ✓System prompt extraction testing
- ✓Credential and secret probing
- ✓Jailbreak and instruction override testing
- ✓Out-of-scope topic probes
- ✓PDF security report with findings
- ✓Remediation guidance for each finding
- ✓Email support
Comprehensive testing across all attack categories with a graded report.
- ✓All 3 tiers (15 attack vectors)
- ✓Multi-turn contextual attacks
- ✓RAG and tool abuse testing
- ✓Indirect prompt injection testing
- ✓Output integrity and bias testing
- ✓Automatic escalation on critical findings
- ✓Graded PDF report (A through F)
- ✓Priority support
Regular testing to keep pace as your AI systems evolve.
- ✓Up to 2 full assessments per week
- ✓All 3 tiers (15 attack vectors)
- ✓Graded PDF report for each assessment
- ✓Ideal for teams shipping frequent AI updates
- ✓Priority support
For specialized use cases, custom engagements, or organizations with unique requirements.
Frequently Asked Questions
What models do you use for testing?
We use leading AI models including Claude (Anthropic) and GPT-4 (OpenAI) for our attack campaigns. The judge model that evaluates responses is always a top-tier model for maximum accuracy. You can also select which model to use for each engagement.
Is this safe to run against production systems?
Yes. Our tests are read-only probes. They send prompts and observe responses. We never attempt to modify data, execute code, or access backend systems. Rate limiting is configurable per engagement so you can control the load on your system.
How long does an assessment take?
A Quick Scan typically completes in 10 to 15 minutes. A Full Assessment takes 30 to 60 minutes depending on rate limits and target response times. You can monitor findings in real time during the assessment.
What do I need to provide to get started?
Just the URL of your chatbot or AI system endpoint, plus any context about what the system is supposed to do and what it should not reveal. We handle everything else.
What if my system passes all tests?
That is a great outcome. You get a clean report with an A grade that you can share with stakeholders, include in compliance documentation, or use to build customer trust. We recommend re-testing periodically as your system evolves.
Can you test internal or pre-production systems?
Yes. For systems not exposed to the public internet, we can work with you on secure access. Contact us and we will find a solution that works for your setup.