FutureSearch DRB - Populate Reference Class: leaderboard
FutureSearch Deep Research Bench task slice for building reference classes from researched examples.
Metric: Average Score. Source: evals.futuresearch.ai. Saturation forecast: Around October 2028. 22 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.6 | 0.4 | #63 |
| 2 | GPT-5.5 | 0.4 | #31 |
| 3 | Gemini 3.1 Pro (Preview) | 0.38 | #58 |
| 4 | Claude Sonnet 4.6 | 0.37 | #90 |
| 5 | Gemini 3 Flash | 0.37 | #99 |
| 6 | Gemini 3 Pro | 0.36 | #80 |
| 7 | Claude Opus 4.5 | 0.36 | #83 |
| 8 | Grok 4 | 0.35 | #181 |
| 9 | O3 | 0.35 | #124 |
| 10 | Claude Opus 4.8 | 0.35 | #37 |
| 11 | Claude Sonnet 4.5 | 0.34 | #142 |
| 12 | GPT-5 | 0.34 | #98 |
| 13 | Claude Opus 4 | 0.33 | #163 |
| 14 | Claude Opus 4.1 | 0.32 | #140 |
| 15 | Claude Sonnet 4 | 0.31 | #199 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=futuresearch-drb-populate-reference-class · How It Works · Data refreshed daily, snapshot 2026-10-20.