Giskard
Giskard offers AI red teaming and testing of agents for security risks (prompt injection, data disclosure) and quality risks (hallucination, refusals), with a paid Hub and an open-source edition.
Visit Giskard Back to comparison
Last checked 9 November 2026. Facts come from Giskard's own public pages, linked below.
At a glance
| Type | AI-assisted testing service |
|---|---|
| Headquarters | France |
| Services listed | AI and LLM penetration testing, Red teaming source |
| Delivery | hybrid Automated black-box red teaming of AI agents, run as a service with human handling of findings, severity and tickets. |
| Scope | Web apps: not published APIs: yes Cloud: not published Network: not published Mobile: not published AI apps: yes |
| Pricing model | mixed |
| Turnaround | Not published No time to first finding or report is stated on the page. source |
| Retesting | Giskard says it re-runs every test until the fix is confirmed. source |
| Sample report | not published |
| Compliance mentioned | HIPAA, GDPR, SOC 2 |
| Integrations | Identity provider, Ticketing (tool not named), CI/CD |
Pricing
| Plan or item | Price | Unit | What it covers |
|---|---|---|---|
| Free | $0 | no billing unit given | Open-source library for solo LLM experiments. source |
Enterprise has no listed price (contact sales). A limited free trial of the Agentic Red Teamer is offered to startups and small teams; terms not given.
Credentials
No credentials with public evidence found.
Our assessment
Editorial judgment by Excession Labs, based on the facts above. Not a rating.
Good fit when
Teams testing text-based AI agents and RAG chatbots that need both security and quality tests and may need on-premise deployment.
Tradeoffs
No enterprise price and no sample report. SOC 2 Type II and HIPAA are self-reported with a trust center link not opened here, so no credential is listed. Text-based agents only.
Sources
- Giskard homepage (checked 9 November 2026)
- Giskard pricing (checked 9 November 2026)
See something wrong or out of date? Send a correction.
Compare with
- ParameterParameter offers AI-agent pentesting for web apps, APIs and infrastructure, needing only a URL and credentials. One run produces executive, customer, auditor…
- PenligentPenligent is a desktop application for AI-assisted penetration testing, published by Future Share LLC. It combines scoping, tools, verification and reporting,…
- PentiPenti (penti.ai, owned by Certypie Inc) sells credit-based agentic AI penetration testing with plans from $20 to $2,000 per month, free unlimited retests, and…
- Terra SecurityTerra Security offers an AI-agent pentesting platform with human oversight through a workspace called TORCH. It covers web apps, AI red teaming, and external…