The AI arena is free today

Open Superagent

Meld

Paper

Progress Over Time

Interactive timeline showing model performance evolution on Meld

State-of-the-art frontier
Open
Proprietary

Meld Leaderboard

1 models
ContextCostLicense
1
Alibaba Cloud / Qwen Team
Alibaba Cloud / Qwen Team
7B——
Notice missing or incorrect data?
About this benchmark

What is Meld?

MELD (Multimodal EmotionLines Dataset) is a multimodal multi-party dataset for emotion recognition in conversations. Contains approximately 13,000 utterances from 1,433 dialogues extracted from the TV series Friends. Each utterance is annotated with emotion (Anger, Disgust, Sadness, Joy, Neutral, Surprise, Fear) and sentiment labels across audio, visual, and textual modalities.

Meld is a multimodal benchmark evaluating models on multimodal, psychology, and creativity tasks. LLM Stats tracks 1 models on this benchmark, scored on a 0–1 scale. The current average is 0.6, with the leader at 0.6.

Compare leaders on the best AI for multimodal, best AI for psychology and best AI for creativity leaderboards.

Current leaders

Qwen2.5-Omni-7B from Alibaba Cloud / Qwen Team currently leads the Meld leaderboard with a score of 0.570 across 1 evaluated AI models.

1Qwen2.5-Omni-7BAlibaba Cloud / Qwen Team57.0%

Source paper

Title
MELD: A Multimodal Multi-Party Dataset for Emotion Recognition in Conversations
Authors
Soujanya Poria, Devamanyu Hazarika, Navonil Majumder, Gautam Naik, and 2 others
Published
Abstract

Emotion recognition in conversations is a challenging task that has recently gained popularity due to its potential applications. Until now, however, a large-scale multimodal multi-party emotional conversational database containing more than two speakers per dialogue was missing. Thus, we propose the Multimodal EmotionLines Dataset (MELD), an extension and enhancement of EmotionLines. MELD contains about 13,000 utterances from 1,433 dialogues from the TV-series Friends. Each utterance is annotated with emotion and sentiment labels, and encompasses audio, visual and textual modalities. We propose several strong multimodal baselines and show the importance of contextual and multimodal information for emotion recognition in conversations. The full dataset is available for use at http:// affective-meld.github.io.

FAQ

Common questions about the Meld benchmark and leaderboard.

What is the Meld benchmark?

MELD (Multimodal EmotionLines Dataset) is a multimodal multi-party dataset for emotion recognition in conversations. Contains approximately 13,000 utterances from 1,433 dialogues extracted from the TV series Friends. Each utterance is annotated with emotion (Anger, Disgust, Sadness, Joy, Neutral, Surprise, Fear) and sentiment labels across audio, visual, and textual modalities.

What is the Meld leaderboard?

The Meld leaderboard ranks 1 AI models based on their performance on this benchmark. Currently, Qwen2.5-Omni-7B by Alibaba Cloud / Qwen Team leads with a score of 0.570. The average score across all models is 0.570.

What is the highest Meld score?

The highest Meld score is 0.570, achieved by Qwen2.5-Omni-7B from Alibaba Cloud / Qwen Team.

How many models are evaluated on Meld?

1 models have been evaluated on the Meld benchmark, with 0 verified results and 1 self-reported results.

Where can I find the Meld paper?

The Meld paper is available at https://arxiv.org/abs/1810.02508. The paper details the methodology, dataset construction, and evaluation criteria.

What categories does Meld cover?

Meld is categorized under multimodal, psychology, and creativity. The benchmark evaluates multimodal models.

What is the best open-source model on Meld?

Qwen2.5-Omni-7B by Alibaba Cloud / Qwen Team is the top-ranked open-source model on Meld, with a score of 0.570 (rank #1).

How recent are the Meld leaderboard results?

The Meld leaderboard was last updated in November 2026 and currently includes 1 evaluated models.