Skip to content

Cyber Models Overview

Cyber Models is the prepaid option for running Deep Agentic Scans. Instead of bringing your own AI provider key (BYOK), you purchase tokens that are credited to your workspace wallet and spent each time you run a Deep Agentic Scan.

This lets you run AI-powered penetration testing without managing your own OpenAI, Anthropic, or OpenRouter account. Ostorlab provisions the AI provider access for you and controls spend automatically.

The biggest advantage is access to state-of-the-art models without the usual safety guardrails that limit offensive security work. As an approved provider for offensive security operations, Ostorlab can run these models at full capability, so your scans are not blocked or degraded by content restrictions that would otherwise refuse legitimate penetration-testing tasks.

Cyber Models vs BYOK

Both options power Deep Agentic Scans. You choose one when configuring the scan.

Cyber Models (Prepaid) BYOK (Bring Your Own Key)
AI provider account Managed by Ostorlab You provide your own
Payment Prepaid tokens from your wallet Billed directly by your provider
Setup Buy a token pack Configure a provider key in Agents > BYOK
Spend control Automatic, per-scan Managed on your provider account

For BYOK, see Generate a BYOK Scan Key and Use your BYOK Scan Key in a Scan Profile.

What is a token?

A token is a unit of prepaid AI spend that can be redeemed for AI usage across providers.

Tokens are held in your workspace wallet. You top the wallet up by purchasing a token pack. When a Cyber Models scan starts, the full token cost of the selected effort is debited from your wallet; when the scan stops or completes, any tokens that were not used are refunded.

Scan Effort

When you run a Deep Agentic Scan with Cyber Models, you select a Scan Effort. The effort determines how many tokens the scan consumes and how deeply it investigates.

Scan Effort Tokens per scan Best for
Core 200 High-confidence findings and faster results
Advanced 500 A deeper investigation with medium-confidence findings
Elite 1,000 Exhaustive testing that surfaces speculative and lower-confidence findings

Your wallet balance determines how many scans you can run. For example, a wallet with 1,000 tokens lets you run:

  • 5 Core scans (200 tokens each), or
  • 2 Advanced scans (500 tokens each), or
  • 1 Elite scan (1,000 tokens).

You can mix efforts in any combination as long as you have enough tokens.

How effort affects results

Scan Effort changes how the scan behaves:

  • Core — prioritizes efficiency and reports only high-confidence risks.
  • Advanced — investigates further and reports medium-confidence risks in addition to high-confidence ones.
  • Elite — runs the most exhaustive investigation and may report speculative or lower-confidence risks for your review.

Higher effort means broader coverage and more potential findings, at the cost of more tokens.

Getting started

  1. Purchase tokens to fund your wallet.
  2. Use prepaid tokens in a scan by selecting Cyber Models and a Scan Effort when creating a Deep Agentic Scan.