XHotpotQA (Gold Supporting Facts) - Exact Match: leaderboard

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

#ModelScore
1GPT-4o35.1
2Llama 3.1 405B Instruct33.76
3GPT-4o Mini29.94
4Gemma 2 27B (IT)28.51
5aya-23-35B25.43
6Qwen 2 72B Instruct24.94
7Gemma 2 9B (IT)24.83
8Llama 3.1 8B Instruct21.62
9Qwen 2 7B Instruct19.67
10aya-23-8B15.32

Interactive version: theaggregate.ai/benchmark?slug=xhotpotqa-gold-supporting-facts-exact-match · How It Works · Data refreshed daily, snapshot 2026-09-26.