Every plan runs the same detection engine. Start free to evaluate your own traffic. Paid adds production volume, your own allow and block phrases, IP threat intelligence, higher rate limits, and one toggle that keeps your traffic out of the shared network. That sharing starts on, and you switch it off in dashboard settings.
Pricing last reviewed: 12 September 2026
Count each text item you submit. A user message plus four retrieved chunks uses five validations; retries add checks. Choose a plan for both monthly volume and peak requests per second.
Agent and retrieval screening · Coverage and workload testing · Security and data sharing
Covers a side project at no cost
No card. Start with your email.
Covers a live app with real users
Instant access • Cancel anytime
Everything in Free, plus:
Covers a team running at volume
Instant access • Cancel anytime
Everything in Starter, plus:
A paid plan renews at the same price each month until you cancel, and there is no overage billing. Requests past your plan's monthly limit come back as an HTTP 429 carrying the date your quota resets, and your traffic flows again on that date or the moment you upgrade.
Every plan runs the same detection engine, starting at $0/month. The free tier gives you 10,000 validations a month. Starter is $29/month for 500,000 validations. Business is $99/month for 1,000,000 validations. No per-request fees, no hidden costs, and no credit card for the free tier.
Yes. SafePrompt gives you 10,000 API validations a month at no cost, on the same detection engine paid plans run: pattern detection, external reference detection and AI validation. No credit card. Confirm your email, then obtain your key in the dashboard. Blocked attacks from free-tier traffic join the shared network defence. Paid plans can turn that contribution off in Privacy Settings.
The free tier includes 10,000 validations a month, the full detection engine (the same one paid plans run), pattern and AI validation, deployment-enabled session tracking without guaranteed conversation isolation, and the shared network defence. Dashboard/API export supports selected records, with record limits and retained hashes excluded. Self-service deletion does not remove retained hashes separately; contact [email protected] for additional access or erasure requests. Paid plans add production volume, IP threat correlation and an intelligence opt-out.
Yes. Confirm your email and obtain your key in the dashboard: 10,000 validations a month, no card and no time limit. The one condition is that attacks blocked on the free plan join the shared network defence; paid plans can switch that off. While sharing is on, a paid account contributes safe and blocked prompts to the intelligence sample store; Free contributes blocked prompts. Request metadata logging and deployment-enabled session storage are separate.
Yes. Upgrade any time, downgrade at the end of any billing cycle. No contracts and no penalties.
Yes. Cancel any time from your dashboard. No cancellation fees and no questions asked.
No card is required. Confirm your email, get your key in the dashboard, then follow the quick start to evaluate your first request.
Every plan runs the same engine: pattern detection, AI validation, deployment-enabled session tracking without guaranteed conversation isolation and the shared network defence.
Yes. Upgrade from your dashboard settings any time. Your new quota is live as soon as payment confirms.
SafePrompt checks the text before it reaches your model, so it works with OpenAI, Anthropic, Google, Mistral and anything else you run. Switch providers any time with no change on your side, and export your data from the dashboard on any plan.
SafePrompt is one API call with nothing to run: no model to host, no GPU to size, no rules to maintain, and a free tier so you can start today. Run LLM Guard self-hosted if your data can never leave your own machines.