Best AI Security Testing Tools in 2026
Updated
In short: PromptGuard is ranked #1 of 28 as of 3 October 2026, ahead of AgentScan and OpenSecureAI Scanner. The best-ranked option with a free plan is AgentScan. The lowest first paid tier on this page is PromptGuard at $19/mo.
28 ai security testing tools ranked on what their makers publish — plans and prices, free tiers, platforms and the facts on their own pages.
| # | Platform | Score | Why | From | Free plan | Paid from | Prompt injection tests | |
|---|---|---|---|---|---|---|---|---|
| 1 | PromptGuard | 8.9 | RecognisedAPIDocumented | $19/mo | Yes | — | Yes | View |
| 2 | AgentScan | 8.6 | RecognisedAPIDocumented | $29/mo | Yes | 29 /mo | Yes | View |
| 3 | OpenSecureAI Scanner | 8.3 | RecognisedAPIDocumented | $49/mo | Yes | 49 /mo | Yes | View |
| 4 | PyRIT | 7.8 | RecognisedAPIDocumented | Free | Yes | — | Yes | View |
| 5 | Project Moonshot | 7.8 | RecognisedAPIDocumented | Free | Yes | — | Yes | View |
| 6 | Akto Agentic Red Teaming | 6.7 | RecognisedAPIDocumented | — | — | — | Yes | View |
| 7 | Promptfoo | 6.7 | RecognisedAPIDocumented | Free | Yes | — | — | View |
| 8 | Crucible | 6.7 | RecognisedAPIDocumented | — | Yes | — | Yes | View |
| 9 | Basilisk | 6.7 | RecognisedAPIDocumented | Free | Yes | — | Yes | View |
| 10 | Mindgard | 6.6 | RecognisedAPIDocumented | — | — | — | Yes | View |
| 11 | AIRTaaS | 6.6 | RecognisedAPIDocumented | — | Yes | — | Yes | View |
| 12 | Giskard | 6.5 | RecognisedAPIDocumented | — | Yes | — | — | View |
| 13 | RedForge | 6.4 | RecognisedAPIDocumented | — | — | — | Yes | View |
| 14 | Prompt Security AI Red Teaming | 6.0 | RecognisedAPIDocumented | — | — | — | Yes | View |
| 15 | AIRTA Red Team | 6.0 | RecognisedAPIDocumented | — | Yes | — | Yes | View |
| 16 | DeepTeam | 5.9 | RecognisedAPIDocumented | — | — | — | Yes | View |
| 17 | garak | 5.8 | RecognisedAPIDocumented | — | Yes | — | — | View |
| 18 | Augustus | 5.8 | RecognisedAPIDocumented | — | — | — | Yes | View |
| 19 | Red Team LLM Lab | 5.6 | RecognisedAPIDocumented | — | Yes | — | Yes | View |
| 20 | SuperRed | 5.6 | RecognisedAPIDocumented | — | Yes | — | Yes | View |
| 21 | SENTINEL RED | 5.4 | RecognisedAPIDocumented | — | — | — | Yes | View |
| 22 | HarmBench | 5.4 | RecognisedAPIDocumented | — | Yes | — | — | View |
| 23 | PromptRAID | 5.2 | RecognisedAPIDocumented | — | — | — | Yes | View |
| 24 | AgentDojo | 5.2 | RecognisedAPIDocumented | — | Yes | — | Yes | View |
| 25 | CyberSecEval | 5.2 | RecognisedAPIDocumented | — | Yes | — | Yes | View |
Is your platform on this list?
Numbered spots on this list can be sponsored. They are labelled, and the editorial order and scores never change for payment.
Questions about this list
Which ai security testing tool is ranked first on Inferse?
PromptGuard is ranked #1 of 28 with a score of 8.9. AgentScan is second and OpenSecureAI Scanner third.
How many of these have a free plan?
6 of the 25 on this page publish a free plan on their own pricing pages.
Which is the cheapest paid option?
On this page, PromptGuard has the lowest first paid tier we found: $19/mo.
How is this list ranked?
Ranked on what each maker publishes, open and connectable first: a public API, open-source code, the depth of its documentation and a free tier to try it on. Model lists are sorted by the figure in their title, exactly as each provider publishes it. Paid placements never change a rank.























