How much does AI red teaming cost?

Most AI red teaming is sold by quote, so there is no market price to report. Of the 23 vendors in this comparison, 2 publish a dollar price, 4 offer a free tier, and no managed-service or crowdsourced vendor publishes a price at all.

Everything below is taken from vendor pricing pages as they read on 2026-11-07. Prices change; the source links are the authority.

Published prices

Free tiers without a public paid price

• SPLX: free plan; Professional (2,000 credits a month) and Enterprise by quote. Also publishes Agentic Radar, a free open-source scanner.
• Giskard: Apache-2.0 open-source library is free; Giskard Hub Enterprise by quote; a limited free trial of the Agentic Red Teamer.

Free assessments or scans with unstated scope

Lakera (a "start for free" button exists; what it covers is not stated); Lasso Security (a "free assessment" is offered; scope not stated); Repello AI (a free red-teaming scan for agents is offered; scope not stated); Straiker (a free ai risk assessment is offered; scope not stated). Treat these as lead-generation offers until the vendor tells you what they include.

Quote-only vendors

15 vendors publish no price and no free tier: Adversa AI (both (platform + service)), Bishop Fox (managed service), Bugcrowd (managed service), Cisco AI Defense (automated platform), F5 AI Red Team (automated platform), Gray Swan (both (platform + service)), HackerOne (managed service), HiddenLayer (automated platform), Holistic AI (automated platform), Mindgard (both (platform + service)), NCC Group (managed service), Noma Security (automated platform), Pillar Security (automated platform), Prisma AIRS (Palo Alto) (automated platform), Trail of Bits (managed service).

Every consultancy and crowdsourced provider in the comparison (Bishop Fox, Bugcrowd, HackerOne, NCC Group, Trail of Bits) is quote-only. The published variables are time, not money: HackerOne engagements run 15 or 30 days; Mindgard managed engagements typically run 4-6 weeks per application; Bugcrowd's AI Pen Test can start within 72 hours and includes a final report and one retest.

What a quote usually depends on

Vendors do not publish their pricing formulas, so this is inferred from how they describe their offerings, not from stated price factors:

• Scope: one chatbot, or several apps, models and agents. Mindgard quotes engagements per application.
• Delivery model: self-serve platform credits (Enkrypt, SPLX) versus a human engagement with a fixed window (HackerOne, Bugcrowd).
• Depth: prompt-level probing versus supply-chain, serialization and hardware review (Trail of Bits describes the latter).
• Reporting: raw findings versus mappings to OWASP, MITRE ATLAS, NIST AI RMF, EU AI Act or ISO 42001 for auditors.
• Deployment: SaaS versus on-prem or private cloud (Giskard Hub, F5 AI Red Team and Promptfoo On-Prem all list this option).
• Continuity: one assessment versus continuous or CI/CD-triggered runs (Straiker, Noma, Promptfoo Enterprise).

Starting at zero

The open-source tools are free: promptfoo (MIT), garak (Apache-2.0), PyRIT (MIT), DeepTeam (Apache-2.0) and Giskard OSS (Apache-2.0). Commercial tiers mostly add dashboards, SSO, continuous monitoring and compliance reporting. See open-source AI red teaming tools.

Common questions

How much does AI red teaming cost?

Most vendors are quote-based. Published figures: Enkrypt AI lists Explore $0, Launch $149/mo ($134 billed annually), Scale $1,499/mo ($1,349 billed annually) and Enterprise custom. Promptfoo has a free Community tier with up to 10k red-team probes a month. SPLX has a free plan and quote-based paid plans. Giskard offers a free open-source library and Enterprise by quote. Open-source tools are free. No managed-service or crowdsourced vendor checked publishes a price.

Which AI red teaming vendors publish prices?

As of 2026-11-07, Enkrypt AI and Promptfoo publish dollar prices. SPLX and Giskard publish a free tier with paid tiers by quote. All other vendors in the comparison are quote-only.

How long does a paid engagement take?

HackerOne runs 15- or 30-day engagements that launch in about a week. Mindgard managed engagements typically run 4-6 weeks per application. Bugcrowd AI Pen Test can start within 72 hours. None of the three publishes a price.