MHGraphBench - Two-hop Selection with Evidence: leaderboard
Metric: Accuracy (%) on the 1,200 two-hop selection items with sanitized PrimeKG feature snippets added, letter-only answers; the OpenAI API models answer at temperature 0 (at most 120 completion tokens) with strict answer-letter parsing, the open models are scored by forced-choice option-letter log-probabilities; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.2 Instant | 67.58 |
| 2 | GPT-5 Mini | 65.42 |
| 3 | GPT-5.1 Instant | 64.42 |
| 4 | GPT-4.1 | 61.5 |
| 5 | GPT-4o | 61.42 |
| 6 | Qwen 2.5 32B Instruct | 50.08 |
| 7 | Qwen 2.5 7B Instruct | 37.75 |
| 8 | DeepSeek R1 Distill Qwen 32B | 35.75 |
| 9 | Llama 3.1 8B Instruct | 31.08 |
| 10 | Mistral 7B Instruct (v0.3) | 28.25 |
| 11 | DeepSeek-R1-Distill-Qwen-7B | 25 |
Interactive version: theaggregate.ai/benchmark?slug=mhgraphbench-two-hop-selection-with-evidence · How It Works · Data refreshed daily, snapshot 2026-10-07.