MHGraphBench - Two-hop Selection with Evidence: leaderboard
Two-hop selection questions with sanitized PrimeKG feature snippets added, in MHGraphBench (2026), a knowledge-graph-grounded mental-health benchmark built from a curated PrimeKG subgraph, with letter-only multiple-choice answers. Accuracy in percent; higher is better.
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 February 2027. 15 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 Instant | 67.58 | #304 |
| 2 | GPT-5 Mini | 65.42 | #364 |
| 3 | GPT-5.1 Instant | 64.42 | #472 |
| 4 | GPT-4.1 | 61.5 | #505 |
| 5 | GPT-4o | 61.42 | #751 |
| 6 | Qwen 2.5 32B Instruct | 50.08 | #1247 |
| 7 | BioMistral-7B | 38.42 | #8332 |
| 8 | Qwen 2.5 7B Instruct | 37.75 | #2723 |
| 9 | Llama3-Med42-8B | 36.17 | #5084 |
| 10 | DeepSeek R1 Distill Qwen 32B | 35.75 | #1834 |
| 11 | Llama 3.1 8B Instruct | 31.08 | #4600 |
| 12 | Mistral 7B Instruct (v0.3) | 28.25 | #6750 |
| 13 | DeepSeek-R1-Distill-Qwen-7B | 25 | #8055 |
| 14 | Llama3-OpenBioLLM-8B | 24.33 | #6364 |
| 15 | Meditron-7B | 22.08 | #8733 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=mhgraphbench-two-hop-selection-with-evidence · How It Works · Data refreshed daily, snapshot 2026-11-21.