HOLMES: leaderboard
Logical reasoning is essential for reliable AI, yet existing benchmarks are largely first-order-logic-centric, focusing on object-level deduction over fixed predicates. This misses many realistic scenarios where models must reason over rules, predicates, functions, constraints, and decision procedures themselves. We introduce HOLMES (Higher-Order Logic Meets real-world Explainable Symbolic reasoni
Metric: Overall Accuracy (self-reported). Source: benchmarklist.com. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.6 Flash | 59.54 | #214 |
| 2 | MiniMax-M2.5 | 54.77 | #303 |
| 3 | DeepSeek V4 Flash | 50.98 | #137 |
| 4 | DeepSeek R1 | 50.11 | #251 |
| 5 | DeepSeek V3.2 | 47.43 | #205 |
| 6 | MiniMax-M2.7 | 46.78 | #258 |
| 7 | GPT-5.4 Mini | 45.39 | #212 |
| 8 | DeepSeek V4 Pro | 44.49 | #102 |
Interactive version: theaggregate.ai/benchmark?slug=holmes · How It Works · Data refreshed daily, snapshot 2026-10-20.