Benchmark
AttuneBench
Evaluation of emotional intelligence in real multi-turn human-AI conversations.
Published leaderboard
Explore scores and evaluation settings on the publisher’s leaderboard.
Open leaderboardCoverage
The paper reports 200 real conversations, participant-provided turn-level annotations, and 11 evaluated models.
Methodology
Compare model predictions and responses against participants' own emotion, behavior, and response-preference annotations; preserve individual metrics and evaluation mode.