G
Garak
LLM Vulnerability Scanner & Generative AI Red-Teaming Kit
Customer Support
Intermediate
About Garak
Garak is an open-source LLM vulnerability scanner developed by NVIDIA. It serves as a generative AI red-teaming and assessment toolkit to systematically probe large language models and dialogue systems for security weaknesses, including hallucination, data leakage, prompt injection, misinformation, toxicity, and other failure modes. Garak helps organizations assess model security, improve robustness, and raise the overall standard of LLM safety. It supports major LLM providers and is actively maintained by NVIDIA and the community.
Pricing
Popular
Open Source
$0 /mo
Use Cases
- Red-teaming and security assessment of LLMs
- Vulnerability scanning before production deployment
- Research on LLM safety and robustness
- Automated probing for prompt injection, hallucination, and toxicity
- Enterprise AI security compliance and auditing
- Improving model guardrails and safety measures
Community Reviews (0)
Sign in to reviewQuick Info
- Website
- garak.ai
- Skill Level
- Intermediate
Categories
Tags
ai-red-teaming
garak
generative-ai-security
llm-security
llm-vulnerability-scanner
nvidia-garak
prompt-injection
Integrations
- GitHub
- Discord
- OpenAI
- Groq
- Hugging Face
- Replicate
- AWS Bedrock
- Cohere
- NVIDIA NIM
- REST APIs
- Local GGUF models
- Command Line Tools
- Python Ecosystem
- Custom LLM Endpoints