specification: API Commons Plans
specificationVersion: '0.1'
schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/Plans
provider: Protect AI
providerId: protectai
created: '2026-06-20'
modified: '2026-06-20'
reconciled: false
tags:
- AI
- ML
- Security
- LLM
- Guardrails
- Plans
description: >-
  Protect AI splits cleanly into two pricing models. The developer-facing LLM
  Guard (and ModelScan) are free, open-source (Apache 2.0) software that you
  self-host with no per-call or license cost - you pay only for the
  infrastructure you run them on. The commercial platform products - Guardian
  (model scanning), Recon (red-teaming), and Layer (runtime) - are sales-led
  with no published list pricing; pricing is negotiated per deployment, now
  also available through Palo Alto Networks as part of the Prisma AIRS
  platform. Specific commercial rates are not reconciled in this artifact.
notes: >-
  Open-source pricing is verifiable from the Apache 2.0 licenses on the
  llm-guard and modelscan repositories. Commercial pricing is not publicly
  listed and requires contacting Protect AI / Palo Alto Networks sales;
  confirm on reconciliation.
sources:
- https://github.com/protectai/llm-guard
- https://github.com/protectai/modelscan
- https://protectai.com/
- https://www.paloaltonetworks.com/prisma/prisma-airs
plans:
- id: protectai-llm-guard-oss
  name: LLM Guard (Open Source)
  type: free
  description: >-
    Self-hosted, Apache 2.0 prompt/output scanner library and FastAPI service.
    Free to use and deploy; you bear only your own compute and operations cost.
  entries:
  - label: LLM Guard Library + API
    name: llm_guard_oss
    type: flat
    metric: license
    limit: -1
    timeFrame: month
    geo: global
    unit: 1
    price: free (Apache 2.0, self-hosted)
    userMultiplied: false
  elements:
  - name: Input Scanners
  - name: Output Scanners
  - name: Self-Hosted llm-guard-api
- id: protectai-modelscan-oss
  name: ModelScan (Open Source)
  type: free
  description: >-
    Self-hosted, Apache 2.0 model serialization-attack scanner; the open-source
    engine behind the commercial Guardian product.
  entries:
  - label: ModelScan CLI / Library
    name: modelscan_oss
    type: flat
    metric: license
    limit: -1
    timeFrame: month
    geo: global
    unit: 1
    price: free (Apache 2.0, self-hosted)
    userMultiplied: false
  elements:
  - name: Pickle / H5 / SavedModel Scanning
- id: protectai-commercial
  name: Commercial Platform (Guardian / Recon / Layer)
  type: enterprise
  description: >-
    Sales-led commercial AI security platform covering model scanning
    (Guardian), automated red-teaming (Recon), and runtime protection (Layer),
    now offered through Palo Alto Networks Prisma AIRS. No public list pricing.
  entries:
  - label: Enterprise Agreement
    name: enterprise
    type: flat
    metric: contract
    limit: -1
    timeFrame: year
    geo: global
    unit: 1
    price: contact sales
    userMultiplied: false
  elements:
  - name: Guardian Model Scanning
  - name: Recon Red-Teaming
  - name: Layer Runtime Protection
  - name: Prisma AIRS Integration
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com