From startups building their first AI agents to enterprise labs developing frontier models, Gray Swan provides flexible pricing that scales with your needs. Start for free and upgrade as your AI security requirements grow.
The flagship integrated solution combining automated testing, real-time protection, and continuous threat intelligence.
What’s included:
Bidirectional protection for AI applications with custom policy enforcement.
Starts at 50 million free tokens, then scales predictably with usage.
Key Features:
Continuous AI security testing integrated with your CI/CD pipeline.
What’s included:
Cash prize competitions and career advancement opportunities
Weekly challenges for AI red-teaming mastery
Dedicated security assessments by expert researchers
Specialized security research for frontier AI capabilities
Agent Shield includes 50M tokens/month with core security policies. Shade and Agent Shield Pro offer free trials to test with your specific deployment. Arena, Proving Ground, and Resource Hub are completely free to use.
Yes, all plans are flexible. You can start with free tiers, upgrade to paid plans as your usage grows, or move between service levels as your needs change.
Yes, we provide custom pricing for high-volume deployments and enterprise packages. Contact our sales team to discuss volume discounts and enterprise agreements.
Support varies by plan—from community support on free tiers to priority support with SLAs on enterprise plans. All customers have access to comprehensive documentation and our active Discord community.
You can start with free tiers immediately. For enterprise deployments, typical onboarding takes 1-2 weeks including initial consultation and custom policy configuration.
No long-term commitments are required for Agent Shield. Custom configurations may require minimum term commitments.
Unlike most subscription-based AI security tools on the market, Agent Shield Core’s industry-leading protection is priced based on token usage, ensuring you only pay for what you use. Pricing for Agent Shield Pro and Shade automated AI red-teaming are based on the complexity of your application and the number of attack behaviors needed to ensure complete coverage, as well as the frequency of test runs.