Frontier AI Test

PySys project for testing various Frontier AI models, currently OpenAI and Anthropic. The project uses the Python client libraries for both models to investigate interactions with them, with a view to extending the project to alignment evaluations. Note that paid subscriptions and API keys are required to run the tests, e.g. the below should be included in your ~/.bashrc file.

export OPENAI_API_KEY=<openai key>
export ANTHROPIC_API_KEY=<anthropic key>

Project Structure

The project layout is as defined below;

frontier-ai-test/
├── README.md
├── pysysproject.xml       # pysys project file
├── src
│   └── utils
│       └── docker.py      # utils for running docker as tool
└── test
    ├── models
    │   ├── anthropic      # tests using anthropic models e.g. claude-opus-4-6
    │   └── openai         # tests using openai models e.g. gpt-4.1
    └── utils
        └── docker         # tests for the docker tools