Selene 1 is a market-leading LLM Judge, available via API. Selene is specifically trained to evaluate AI responses, and outperforms frontier models from leading labs—including OpenAI's o-series, Anthropic's Claude 3.5 Sonnet, and DeepSeek's R1—across 11 commonly used benchmarks for evaluators.

Selene excels at a variety of evaluation tasks including absolute scoring, classification, and pairwise preference. It provides sores and actionable chain-of-thought critiques to improve GenAI performance.

The Eval Copilot (formerly Alignment Platform) complements Selene by helping users automatically generate, test, and refine custom evaluation metrics with minimal prompt engineering. Simply describe your eval task, and the platform builds evaluation criteria tailored to your specific use case.

The Selene API can be used in diverse applications with or without reference responses—detecting hallucinations in RAG systems, assessing logical reasoning in agents, verifying correctness in specific domains, and much more. Users can leverage Selene to iteratively test and refine prompts, objectively compare different models' performance, and continuously monitor AI systems in production environments.

Start evaluating your GenAI products with Selene and ship with confidence.