Kaggle FACTS Search: leaderboard
FACTS Search sub-task evaluating factual accuracy when models can use search/retrieval to augment their responses.
Metric: Score (%). Source: www.kaggle.com. Saturation forecast: Around January 2027. 29 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 85.56 | #58 |
| 2 | GPT-5.5 | 85.02 | #31 |
| 3 | Gemini 3 Flash (Preview) | 81.04 | #82 |
| 4 | Claude Opus 4.8 | 80.29 | #37 |
| 5 | Claude Opus 4.7 | 78.8 | #49 |
| 6 | GPT-5 | 77.71 | #98 |
| 7 | Grok 4.1 Fast (Reasoning) | 77.44 | #217 (Grok 4.1 Fast) |
| 8 | Claude Opus 4.6 | 76.7 | #63 |
| 9 | Claude Sonnet 4.6 | 76.42 | #90 |
| 10 | Grok 4 | 75.32 | #181 |
| 11 | O3 | 74.79 | #124 |
| 12 | Claude Opus 4.5 | 73.25 | #83 |
| 13 | GPT-5.2 | 72.16 | #110 |
| 14 | Claude Sonnet 4.5 (Thinking) | 69.8 | #142 (Claude Sonnet 4.5) |
| 15 | Gemma 4 31B | 68.88 | #200 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=kaggle-facts-search · How It Works · Data refreshed daily, snapshot 2026-10-20.