RuleWorld - Multi-Hop (Natural Language): leaderboard
Metric: Exact-match accuracy (%; x100 of the 0-1 score; multi-hop rule QA with 2 to 4 hops, mean of the three hop counts, rules in natural language; all 1,000 abstract, non-commonsense procedural rules of the pool prepended to the prompt (full-context prompting); exact match on the boxed answer list, partial credit per matching position; 10 questions per task type and difficulty level for each of five seeds; temperature 0 (Qwen3-32B 0.6)). Source: arxiv.org. Saturation forecast: Around March 2027. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6 | 64.33 |
| 2 | GPT-5.5 | 46.67 |
| 3 | DeepSeek V3.2 (Non-reasoning) | 38 |
| 4 | Qwen 3 32B | 33.33 |
| 5 | GPT-5.1 | 28 |
| 6 | Qwen 2.5 32B Instruct | 22 |
| 7 | Qwen 2.5 14B Instruct | 19.33 |
| 8 | Llama 3.3 70B Instruct | 18.67 |
| 9 | Qwen 2.5 72B Instruct | 18.67 |
| 10 | Llama 3.1 70B Instruct | 14.67 |
Interactive version: theaggregate.ai/benchmark?slug=ruleworld-multi-hop-natural-language · How It Works · Data refreshed daily, snapshot 2026-09-26.