Delulu - Parameter Hallucination Detection: leaderboard

LLMs as judges on the 435 Delulu (2026) fill-in-the-middle samples whose hallucinated completion contains parameters that do not exist: the judge must accept the real completion and reject the hallucinated one. The score is the both-correct rate (%); higher is better.

Metric: Both-correct rate (%) on the 435 Delulu samples whose hallucination is parameters that do not exist, as an LLM judge: the judge sees each sample twice, once with the gold completion and once with the hallucinated one, and must accept the first and reject the second (both-correct rate, %), with chain-of-thought and a binary verdict; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Claude Opus 4.591.1#160
2Claude Sonnet 4.688.5#173
3GPT-5.2 Codex87.2#182
4GPT-5.479.8#135
5Claude Haiku 4.574.3#583
6GPT-5.4 Mini66.3#406
7GPT-4.1 Mini63.5#789
8GPT-4o Mini46.1#1667

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

Interactive version: theaggregate.ai/benchmark?slug=delulu-parameter-hallucination-detection · How It Works · Data refreshed daily, snapshot 2026-11-21.