← Back to changelog

October 7, 2026

Evaluate with OpenAI decision models

Tobias Wochinger

Use OpenAI's gpt-6-luna decision model to score observations with choice, score, and yes/no questions.

You can now use OpenAI's gpt-6-luna in decision-model evaluators. Select an existing OpenAI connection, ask choice, score, or yes/no questions, and write one Langfuse score per answer.

OpenAI decision models evaluate the observation input through the OpenAI Decisions API. This makes them useful for high-volume classification, rubric scoring, and binary checks alongside TypeSafe Jev.