DeepResearch Bench - Instruction Following — leaderboard
Metric: Score (%). Source: huggingface.co. 45 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 50.12 |
| 2 | Claude 3.7 Sonnet | 44.05 |
| 3 | Sonar Pro | 43.39 |
| 4 | Gemini 2.5 Pro (Preview 05-06) | 40.24 |
Interactive version: theaggregate.ai/benchmark?slug=deepresearch-bench-instruction-following · How the rankings work · Data refreshed daily, snapshot 2026-07-22.