One platform, eleven classes, and what each of the others publishes beside it

Pentera

As published

Core publishes "OWASP Top 10 Testing" as a shipping capability (pentera.io/pentera-core/, read 2026-09-14). A separate AI-native web application capability was announced 29 Jul 2026 "in beta to selected customers, with general availability rolling out in Q4 2026" (pentera.io/press-release/, read 2026-09-14; re-read 2026-10-28: no GA announcement published yet).

First-class on the platform?

Yes: shipping in Core; the AI-native successor is in beta.

XBOW

As published

"XBOW currently supports testing web applications and their APIs" (docs.xbow.com/console/reference/asset-types, read 2026-10-28).

First-class on the platform?

Yes: the core of the product.

Hadrian

As published

Nova: "Point Nova at an application"; "a fleet of AI hacker agents works through it the way a skilled pentester would, step by step, until it can show real impact"; most tests complete in 24–48 hours; from €1,250 per test ex-VAT. No formal target-type list is published.

First-class on the platform?

Yes, it is Nova’s entire job, without a published target taxonomy.

B-52

As published

Covered, including authenticated flows and the authorisation decisions behind each role, credential supplied per role; reachable from the self-serve flow (b52security.com/platform/, read 2026-10-28).

First-class on the platform?

Yes: first-class on the platform.

B-52

As published

Covered on the platform, all three delivery models, reachable from the self-serve flow. Release builds tested as published: obfuscated, certificate-pinned, defended at runtime, no source required.

First-class on the platform?

Yes: first-class on the platform.

Pentera

As published

Surface names APIs as an asset class: "Validate VPNs, Git, SSH, storage, APIs, and web services — not just web applications" (pentera.io/pentera-surface/, read 2026-09-14).

First-class on the platform?

Yes: named asset class on Surface.

XBOW

As published

Supported, in the same sentence as web applications (docs.xbow.com/console/reference/asset-types, read 2026-10-28).

First-class on the platform?

Yes.

B-52

As published

Covered as its own class: an authorisation matrix written first, then routes checked with a second credential, reachable from the self-serve flow.

First-class on the platform?

Yes: first-class on the platform.

B-52

As published

Covered on the platform, reachable from the self-serve flow. Four application types worked: Windows native (.NET, C++, Win32), Java desktop, Electron and web-wrapped, and client-server protocol/local-storage analysis.

First-class on the platform?

Yes: first-class on the platform.

Pentera

As published

Pentera Surface: "External network security validation" (pentera.io/pentera-platform/, read 2026-09-14).

First-class on the platform?

Yes: a named module.

Horizon3

As published

External Pentesting, in every tier including entry (horizon3.ai/pricing/, read 2026-10-28); executed in Horizon3’s cloud on "dedicated, ephemeral resources" (horizon3.ai/nodezero/, read 2026-09-14).

First-class on the platform?

Yes.

Hadrian

As published

Atlas: "Continuous external exposure management": "Everything your company exposes on the internet," discovered "from the outside," domains, DNS, IPs, certificates, cloud services and apps; exposures "chained and exploited the way a skilled attacker would".

First-class on the platform?

Yes, as discovery plus exposure validation; pentest depth sits in Nova.

B-52

As published

Covered: perimeter worked out from announced ranges, resolving names and what answers on each address, then proposed back to you as the scope.

First-class on the platform?

Yes: first-class on the platform.

Pentera

As published

Pentera Core: "AI-driven pentesting to continuously test your production environments at scale": lateral movement, privilege escalation, domain admin compromise, credential assessment, segmentation (pentera.io/pentera-core/, read 2026-09-14).

First-class on the platform?

Yes, and deeper than B-52. Conceded by name; see concessions.

Horizon3

As published

Internal Pentesting in every tier; runs "from a free Docker host or open virtualization appliance (OVA) that you can set up in minutes" (horizon3.ai/nodezero/, read 2026-09-14).

First-class on the platform?

Yes: their home ground.

B-52

As published

Covered. The one class that needs something deployed inside your network, with on-premise and customer-VPC deployments available, residency in India, the EU, the US and Singapore. No depth claim against Core anywhere on this estate.

First-class on the platform?

Yes: first-class on the platform.

Pentera

As published

Pentera Cloud: "Cloud identity and hybrid environment security validation" across AWS and Azure (pentera.io/pentera-cloud/, read 2026-09-14).

First-class on the platform?

Yes: a named module.

Horizon3

As published

Cloud Pentesting in every tier; names AWS, Azure and Kubernetes; "NodeZero is deployable both on-prem and in the cloud" (horizon3.ai/nodezero/cloud-pentesting/, read 2026-09-14).

First-class on the platform?

Yes.

Hadrian

As published

Atlas covers "risky cloud services and storage" as an exposure and misconfiguration solution, found externally. Cloud penetration testing as a class is not published.

First-class on the platform?

Partial as published: exposure discovery from the outside, not an inside-the-account pentest class.

B-52

As published

Covered: the identity and permission layer, proved by making the call.

First-class on the platform?

Yes: first-class on the platform.

Pentera

As published

Core: credential and password assessment, domain admin compromise (pentera.io/pentera-core/, read 2026-09-14).

First-class on the platform?

Yes, and deeper than B-52. Conceded by name; see concessions.

Horizon3

As published

"AD password audits" in every tier (horizon3.ai/pricing/, read 2026-10-28); internal pentesting page names credential attacks, password spraying, Zerologon, aligned to MITRE ATT&CK (horizon3.ai/nodezero/internal-pentesting/, read 2026-09-14).

First-class on the platform?

Yes, as an audit inside every tier.

B-52

As published

Covered: privilege paths walked end to end. Not to the depth Pentera Core reaches; that concession lives on /compare/pentera/ and is not repeated defensively here.

First-class on the platform?

Yes: first-class on the platform.

Pentera

As published

"Phishing Attack Emulation — Emulate phishing campaigns to determine whether attackers can capture credentials or execute malicious payloads" (pentera.io/pentera-surface/, read 2026-09-14).

First-class on the platform?

Yes, in the product, on Surface.

Horizon3

As published

Phishing Impact Testing in every tier (horizon3.ai/pricing/, read 2026-10-28).

First-class on the platform?

Yes.

B-52

As published

Covered, scoped separately from the other classes. Three forms, one boundary: targeted phishing simulation; pretexting and credential harvesting inside an attack chain; OSINT and target enumeration feeding the campaign. Run under written authorisation naming the population and the window.

First-class on the platform?

Yes: first-class on the platform.

B-52

As published

Covered, reachable from the self-serve flow. A file, a line, the path an attacker takes to reach that line, and the closing change in the repository’s own idiom. Code deleted after thirty days of inactivity.

First-class on the platform?

Yes: first-class on the platform.

B-52

As published

Covered on the platform. Three forms: prompt injection and jailbreak testing (direct and indirect), model endpoint and API security, and agentic systems: tool misuse, autonomy boundaries, unintended chaining. Model provenance and training data deliberately out of scope, by number, on the class page.

First-class on the platform?

Yes: first-class on the platform.