Azure/azure-sdk-for-python azure-ai-evaluation_1.14.0

on GitHub

1.14.0 (2026-01-05)

Bugs Fixed

• Updated CodeVulnerability and UngroundedAttributes evaluators for RedTeam to use the binary true/false scoring pattern so their results align with service responses.
• Fixed GroundednessEvaluator with query not honoring is_reasoning_model (and credential) when reloading the query prompty, which could cause max_tokens to be sent to reasoning models. #44385