Data Agent Benchmark: leaderboard
UC Berkeley EPIC benchmark for data agents answering complex real-world data tasks across datasets, domains, and tool workflows.
Metric: Pass@1 (%). Source: ucbepic.github.io. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.6 | 55.51 | #63 |
| 2 | Gemini 3 Pro | 46.63 | #80 |
| 3 | GPT-5 Mini | 36.63 | #186 |
| 4 | GPT-5.2 | 29.91 | #110 |
| 5 | Gemini 2.5 Flash | 10.49 | #245 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=data-agent-benchmark · How It Works · Data refreshed daily, snapshot 2026-10-20.