Question 1

What is NullVector LLM Scanner?

Accepted Answer

NullVector LLM Scanner is an LLM security scanning platform built to automatically probe your AI models and API endpoints for jailbreaks, prompt injection, data exfiltration, PII leakage, malware generation, and compliance gaps.

Question 2

How does the scanner work?

Accepted Answer

NullVector LLM Scanner runs adversarial probes against your configured targets. Each probe sends a series of crafted prompts to your model, analyzes the responses using specialized detectors, and computes an Attack Success Rate (ASR). Results are stored as structured findings with severity ratings and remediation guidance.

Question 3

What target types are supported?

Accepted Answer

NullVector LLM Scanner supports OpenAI-compatible REST API endpoints, direct OpenAI models, Anthropic / Claude models, custom web chat interfaces (browser-automated via Chromium), and any HTTP endpoint that accepts a chat completion payload.

Question 4

What is an Attack Success Rate (ASR)?

Accepted Answer

ASR is the percentage of adversarial probe attempts that successfully bypassed your model's safety guardrails. A score of 0% means the model resisted all attacks; a score above 30% is considered high risk and warrants immediate remediation.

Question 5

Is there a free plan?

Accepted Answer

Yes. The Free tier gives you 3 scans per month, 1 target, and access to basic probe categories — no credit card required.

Question 6

What is included in the Starter plan?

Accepted Answer

Starter ($19/month or $190/year) includes 25 scans per month, up to 5 targets, no API keys, 30-day report retention, and access to basic probe categories.

Question 7

What does the Pro plan add?

Accepted Answer

Pro ($79/month or $790/year) expands to 250 scans per month, 25 targets, 10 API keys, 90-day report retention, and unlocks jailbreak probes, advanced probes, and PII extraction probes.

Question 8

What does Advanced include?

Accepted Answer

Advanced ($349/month or $3,490/year) provides unlimited scans, unlimited targets, unlimited API keys, 365-day report retention, and all probe categories including compliance probes (GDPR, HIPAA), adversarial ML attacks, and malware generation probes.

Question 9

Does NullVector LLM Scanner store my model's API keys?

Accepted Answer

API keys for your targets are stored encrypted at rest in our database and are never logged, transmitted in plaintext, or shared with third parties. Keys are only used at scan execution time to authenticate against your configured endpoint.

Question 10

What probe IDs are available?

Accepted Answer

NullVector exposes 14 probes across 5 categories: Basic (prompt_injection, data_exfiltration), Pro (dan, role_play, encoding_bypass, context_manipulation, hallucination, pii_extraction), Advanced (compliance_gdpr, compliance_hipaa, adversarial_ml, credential_harvesting, malware_gen, identity_boundary).

How can we help?

Product

Plans & Billing

Security & Privacy

Probe Reference

Technical Reference — Web Chatbot Targets

Still have questions?