Verify LLM API

VerifyLLMAPI provides a robust method for confirming the identity of the underlying language model used by a digital agent or API. Key features include: • Agent-driven model verification • Direct testing of OpenAI/Anthropic endpoints • Reuses host login for fresh sessions • Samples one-token outputs for distribution matching • Local processing of credentials and samples This tool helps users ensure the model behind their agent matches its claimed identity. It operates by analyzing the statistical distribution of single-token outputs and comparing them against a database of public model fingerprints using Jensen-Shannon distance. The process is designed for security, with all sensitive data like credentials and raw samples remaining local to the user's environment. It avoids credential extraction and environment-model guesswork. The library supports verification for 161 exact model IDs, providing conclusive matches or returning 'INCONCLUSIVE' for unknown or weak correspondences. This ensures transparency and trust in the digital agents being utilized. The verification is performed via a default complete check involving 20 fresh model requests across multiple probe cells. Ideal for developers, researchers, and organizations who need to validate the authenticity and consistency of language models integrated into their systems or digital agents, without compromising security or privacy.