DialToM - Prospective (Emotional Support): leaderboard

Prospective theory of mind on DialToM (2026), naturalistic human-human dialogues from emotional support conversations (ESConv): the model must forecast which next dialogue trajectory fits a given mental-state profile, without the dialogue context, choosing among four annotator-verified options. Correctness rate in percent; higher is better.

Metric: Correctness rate (%) on four-option multiple-choice questions of the DialToM Prospective task: forecast the state-consistent next dialogue trajectory from an isolated mental-state profile with no dialogue context (the State-Driven Diagnostic Probe), emotional support domain (naturalistic human-human dialogues from ESConv; GPT-4o-drafted options verified by professional annotators and Dawid-Skene aggregation), zero-shot; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 14 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3 Pro (Preview)80.46#119
2Gemini 2.5 Pro29.06#276
3Qwen 3 235B A22B29.06#628
4Kimi K225.62#443
5Gemini 2.5 Flash25.12#463
6Llama 3.3 70B19.7#1514
7Llama 3.1 8B16.75#5196
8GPT-516.26#178
9GPT-OSS-120B15.76#704
10Mistral Nemo14.29#4484
11GPT-4.112.81#505
12Llama 4 Maverick11.82#1110
13DialToM DeepSeek v3 (checkpoint unspecified)11.82
14Mistral Small 3.24.93#1641

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=dialtom-prospective-emotional-support · How It Works · Data refreshed daily, snapshot 2026-11-21.