Senthex
Employees use ChatGPT.Your CISO has no visibility. EU AI Act says fix it
Last verified:
What is Senthex?
Senthex is an AI‑specific firewall that sits directly between your application and major LLM providers such as OpenAI, Anthropic, Mistral, Gemini, and OpenRouter. It intercepts every LLM API call with essentially one‑line integration—by simply changing the base URL—while leaving your existing OpenAI SDK, API key, and code intact. The service runs 26 parallel shields on every request to detect threats like prompt injections, PII and secret leaks, unicode steganography, behavioral anomalies, and semantic hijacking, blocking or flagging risky queries before they reach the model.
On the operational side, Senthex provides real‑time observability so you can see every request classified as PASS, WARN, or BLOCK along with latency and threat reasons. It includes a live feed, per‑project analytics, and anomaly detection, plus one‑click audit‑trail exports tailored for EU AI Act compliance. The platform is built for production use, with near‑zero latency overhead, no data retention by default, and EU‑hosted infrastructure that is GDPR‑aligned and ready for the upcoming EU AI Act deadline.
Senthex is aimed at startups, scale‑ups, and teams shipping production LLM‑powered apps who want to harden their LLM stack without changing their client‑side code or provider. It appeals to organizations that need both runtime protection and compliance‑ready logging, especially those under EU regulation or planning to operate in Europe. The product is positioned as an independent, European‑built alternative to recently acquired LLM security tools, with a focus on simplicity, full OWASP LLM Top 10 coverage, and startup‑friendly pricing.
Senthex pricing
Pricing model: Freemium
Senthex offers a Free plan with 1,000 requests per month, no credit card required, including all 26 shields, a real‑time dashboard, 3‑day log retention, and community support. The Pro plan is $49/month with 50,000 requests, longer 14‑day log retention, webhook alerts, CSV export, and email support. The Business plan is $199/month with 500,000 requests, 90‑day log retention, EU AI Act audit trail, Webhook + Slack alerts, priority support, and an OWASP LLM Top 10 report. Enterprise starts at $999/month with unlimited requests, unlimited log retention, custom shields, full compliance suite, dedicated support, SLA guarantee, and self‑hosted deployment options.
Senthex pros
- One‑line integration by swapping base_url only
- Wraps multiple LLM providers in a single proxy (OpenAI, Anthropic, Mistral, Gemini, OpenRouter)
- No SDK or code refactor needed on existing apps
- 26 parallel shields covering prompt injection, PII, secrets, unicode, behavioral, semantic, and other threats
- Built‑in OWASP LLM Top 10 coverage
- Near‑zero latency overhead in production
- Real‑time live feed showing PASS, WARN, BLOCK decisions per request
- Per‑project analytics and anomaly detection
- Full audit‑trail exports for EU AI Act compliance
- Zero data retention by default (only metadata logged)
- EU‑hosted and GDPR‑aligned, hosted in Germany
- Compliance‑ready features such as EU AI Act audit trail and OWASP LLM Top 10 reports
- Self‑hosted option available on Enterprise plans
- Supports webhook and Slack alerts across paid tiers
- Scalable from 1K free requests to unlimited Enterprise usage
- Transparent feature parity: all plans include all 26 shields
- Flexible billing: upgrade or downgrade anytime, no time‑boxed trial
- CSV export of logs for local analysis
- Async multi‑turn tracking and behavioral fingerprinting
- Canary token support for detecting and tracing attacks
Senthex cons
- Free tier limited to 1,000 requests per month
- Paid plans may be expensive for very small or experimental projects
- No granular per‑shield billing; you pay for the full shield set regardless of usage
- Enterprise‑level features require custom contracts and negotiation
- Self‑hosted option only available on Enterprise, which has a high entry price
- Limited or no phone support on lower tiers
- Requests above plan limits are blocked rather than queued or throttled
- Dashboard and features may feel complex for non‑technical stakeholders
- Still relatively new as a category‑specific firewall, with an evolving shield set
- No built‑in custom model training or fine‑tuning inside the firewall layer
Frequently asked questions about Senthex
How does the free plan work?
The Free plan includes 1,000 LLM API requests per month with no credit card required. Every request is processed through all 26 shields, and you get full access to the real‑time dashboard, 3‑day log retention, and community support. The request counter resets on the first of each month, effectively acting as an open‑ended trial you can keep using while prototyping or lightly testing.
Can I change plans at any time?
Yes, you can upgrade, downgrade, or cancel your plan at any time from your billing settings. Upgrades take effect immediately and give you the new request limit and features right away. Downgrades apply at the end of your current billing cycle, so you continue using the higher tier until that period ends.
What counts as a request?
One request is defined as a single LLM API call proxied through Senthex, regardless of how many tokens or turns are involved. A streaming response that would normally be one API call from your app still counts as exactly one request, even if the model emits many tokens over time.
What happens when I hit my request limit?
Requests exceeding your monthly plan limit are blocked and return a 429 status code. You can upgrade your plan at any time to increase your request cap, and requests will resume as long as they stay within the new quota. Until then, overage calls are not charged but are simply rejected.
Do you store my request content?
By default, Senthex does not store the raw request and response bodies. It only logs metadata such as timestamps, shield scores, threat classifications, and routing information. If you need deeper inspection or archival, that is handled through explicit retention settings on higher tiers, not automatic long‑term storage of payloads.
Is there a trial for paid plans?
The Free plan effectively serves as a trial: it gives you 1,000 requests per month with all 26 shields, the dashboard, and 3‑day log retention. From there you can upgrade to Pro (50K requests) or higher tiers to unlock full feature access while keeping the same one‑line integration and workflow.
How many LLM providers does Senthex support?
Senthex currently supports OpenAI, Anthropic, Mistral, Gemini, and OpenRouter as first‑class proxied providers. Your application can keep using their respective SDKs and APIs; Senthex injects itself as a universal proxy layer by changing the base URL, so the same Senthex instance can route and protect calls to multiple providers.
Is Senthex compliant with the EU AI Act?
Senthex is designed to be EU AI Act ready, with features such as audit‑trail exports, data classification, and compliance‑oriented logging built into higher‑tier plans. The service is hosted in Germany and runs with zero data retention by default, helping organizations meet key obligations around transparency, logging, and data handling for production LLM workloads.
What is the difference between Pro and Business plans?
The Pro plan offers 50,000 requests per month, 14‑day log retention, webhook alerts, CSV export, and email support, while the Business plan increases the cap to 500,000 requests, extends retention to 90 days, adds EU AI Act audit trail, Webhook + Slack alerts, priority support, and an OWASP LLM Top 10 report. Business is tailored for teams with heavier production traffic and stricter compliance needs.
Can I self‑host Senthex?
Self‑hosted deployment is available only on the Enterprise plan. Customers on Enterprise can run Senthex on their own infrastructure while still benefiting from all 26 shields, custom shield logic, unlimited retention, and SLA‑backed support, giving them full control over where logs and traffic reside.