# FriendBench: Benchmarking Dyadic Familiarity Inference in Humans and Multimodal Large Language Models

arXiv: https://arxiv.org/abs/2607.29602 · HTML: https://mlsanity.com/paper/2607.29602/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in MLLM training and visual instruction tuning, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 10 of its 51 references that are in the map.
- First posted on arXiv 31 Jul 2026.

## Placed in

- Multimodal LLMs / vision-language models / [MLLM training and visual instruction tuning](https://mlsanity.com/mllm/mllm-training-and-visual-instruction-tuning/)

## Builds on (most cited first, 10 of 10)

- [Calibrate Before Use: Improving Few-Shot Performance of Language Models](https://mlsanity.com/paper/2102.09690/) (2021)
- [Clever Hans or Neural Theory of Mind? Stress Testing Social Reasoning in Large Language Models](https://mlsanity.com/paper/2305.14763/) (2023)
- [Context-Aware Personality Inference in Dyadic Scenarios: Introducing the UDIVA Dataset](https://mlsanity.com/paper/2012.14259/) (2020)
- [Dual-Glance Model for Deciphering Social Relationships](https://mlsanity.com/paper/1708.00634/) (2017)
- [A Domain Based Approach to Social Relation Recognition](https://mlsanity.com/paper/1704.06456/) (2017)
- [SIV-Bench: A Video Benchmark for Social Interaction Understanding and Reasoning](https://mlsanity.com/paper/2506.05425/) (2025)
- [DDRel: A New Dataset for Interpersonal Relation Classification in Dyadic Dialogues](https://mlsanity.com/paper/2012.02553/) (2020)
- [HumanSense: From Multimodal Perception to Empathetic Context-Aware Responses through Reasoning MLLMs](https://mlsanity.com/paper/2508.10576/) (2025)
- [Are they lovers or friends? Evaluating LLMs' Social Reasoning in English and Korean Dialogues](https://mlsanity.com/paper/2510.19028/) (2025)
- [AsyReC: A Multimodal Graph-based Framework for Spatio-Temporal Asymmetric Dyadic Relationship Classification](https://mlsanity.com/paper/2504.05030/) (2025)

## Abstract (arXiv)

Reading a social situation often depends on behavior, not words alone. We introduce FriendBench, a benchmark for inferring whether two people are already familiar or are meeting as strangers, from a 20-second clip of a dyadic ice-breaker conversation. Every pair answers the same type of prompt, so only the manner of interaction can reveal the answer. Across text, audio, and video, we compare 26 models from seven companies against matched human panels over 96 balanced dyads. The best model and the human crowd are statistically indistinguishable on accuracy in every modality, but reach it differently: humans stay balanced across the two answers, while the strongest models favor ``stranger.'' This is a difference in effective prior, not in discrimination. Richer channels help both unequally, and only humans gain from visible behavior on top of speech. We release the stimuli, human ratings, and model predictions.