Kaggle DeepSearchQA (Google) — leaderboard
Google DeepMind benchmark for deep research agents on 900 multi-step web information-seeking tasks requiring exhaustive answer sets, entity resolution, and stopping-criteria judgment.
Metric: F1 Score (%). Source: www.kaggle.com. Status: saturation imminent. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 Pro | 78.98 |
| 2 | GPT-5.4 | 78.63 |
| 3 | Gemini 3.1 Pro (Preview) | 75.77 |
| 4 | GPT-5 | 73.24 |
| 5 | Claude Sonnet 4.6 | 56.52 |
| 6 | Gemini 2.5 Pro | 48 |
| 7 | Gemini 2.5 Flash | 42.99 |
| 8 | Claude Opus 4.5 (Thinking) | 40.2 |
| 9 | Claude Sonnet 4.5 (Thinking) | 27.85 |
| 10 | Claude Haiku 4.5 (Thinking) | 22.24 |
Interactive version: theaggregate.ai/benchmark?slug=kaggle-deepsearchqa-google · How the rankings work · Data refreshed daily, snapshot 2026-07-22.