OmniToM - Belief Extraction - False Belief Task: leaderboard

OmniToM (2026) evaluates theory of mind through explicit belief modeling on 895 short social stories from ToMBench, with gold belief propositions built by a human-calibrated annotation pipeline. This board is Stage 1 belief extraction on the 97 False Belief Task stories, macro F1 against the gold belief propositions; higher is better.

Metric: Belief-extraction macro F1 (%): story-level F1 between the extracted world-fact and actor-belief propositions and the gold propositions, matched semantically by a GPT-5 judge, macro-averaged over stories, on the 97 False Belief Task stories, on the 895 OmniToM stories from seven ToMBench categories (22,343 gold belief propositions), zero-shot TELeR level 3 prompts; GPT-5 is not scored as it is the judge; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemma 3 27B (IT)72.39#1324
2Mistral Large 2 (Jul)71.28#1144
3Llama 3.3 70B Instruct64.07#1385
4Qwen 3 32B57.32#972
5Gemini 2.5 Flash56.48#463
6Mistral Small 354.58#1796
7Qwen 3 8B50.22#1873
8Llama 3.1 8B Instruct48.29#4600

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

Interactive version: theaggregate.ai/benchmark?slug=omnitom-belief-extraction-false-belief-task · How It Works · Data refreshed daily, snapshot 2026-11-21.