XHotpotQA (Gold Supporting Facts) - Exact Match: leaderboard
XHotpotQA (2026) tests cross-lingual multi-hop question answering in which the question, bridge evidence and answer evidence may be in different languages. A one-shot reader receives the annotated supporting sentences of 7,405 validation items and is scored by answer exact match in percent under the legacy Hotpot normalization; higher is better.
Metric: Answer exact match (%; one-shot reader given the question and the annotated supporting sentences of the audited V1 validation set, 7,405 mixed-language HotpotQA-derived items, legacy Hotpot answer normalization). Source: arxiv.org. Saturation forecast: Estimated already saturated. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 35.1 | #741 |
| 2 | Llama 3.1 405B Instruct | 33.76 | #1114 |
| 3 | GPT-4o Mini | 29.94 | #1627 |
| 4 | Gemma 2 27B (IT) | 28.51 | #1815 |
| 5 | aya-23-35B | 25.43 | #2669 |
| 6 | Qwen 2 72B Instruct | 24.94 | #1489 |
| 7 | Gemma 2 9B (IT) | 24.83 | #2275 |
| 8 | Llama 3.1 8B Instruct | 21.62 | #4277 |
| 9 | Qwen 2 7B Instruct | 19.67 | #3238 |
| 10 | aya-23-8B | 15.32 | #6778 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=xhotpotqa-gold-supporting-facts-exact-match · How It Works · Data refreshed daily, snapshot 2026-11-12.