LogiQAte - Obfus FOL: leaderboard
First-order-logic entailment questions whose premises are rewritten into logically equivalent but less familiar forms and checked with the Prover9 theorem prover; LogiQAte (Borah et al., 2026). Exact-match accuracy in percent on the obfuscated questions, zero-shot at temperature 0; higher is better.
Metric: Exact-match accuracy (%; first-order-logic entailment questions whose premises are rewritten into logically equivalent forms, zero-shot, temperature 0). Source: arxiv.org. Saturation forecast: Not forecast. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 74 | #741 |
| 2 | Gemma 3 27B (IT) | 71 | #1306 |
| 3 | Claude 3.7 Sonnet | 69 | #455 |
| 4 | Gemini 2.5 Pro | 68 | #262 |
| 5 | O4 Mini | 68 | #354 |
| 6 | Claude Haiku 4.5 | 67 | #568 |
| 7 | QwQ-32B | 61 | #909 |
| 8 | GPT-5 | 56 | #163 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=logiqate-obfus-fol · How It Works · Data refreshed daily, snapshot 2026-11-12.