ADBench: leaderboard
Real-world advertising analytics benchmark for LLM agents, using 100 tasks, 11 domain tools, three difficulty levels, and trajectory-aware evaluation.
Metric: Pass@3 (imported). Source: github.com. Source: benchmarklist.com. Saturation forecast: Around April 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro | 83 | #80 |
| 2 | GPT-5.1 | 82 | #135 |
| 3 | O3 | 82 | #124 |
| 4 | GLM-4.7 | 81 | #192 |
| 5 | DeepSeek V3 | 80 | #321 |
| 6 | Kimi K2 | 79 | #243 |
| 7 | Qwen 3 235B A22B | 68 | #313 |
| 8 | Qwen 3 32B | 59 | #443 |
| 9 | Qwen 3 8B | 58 | #674 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=adbench · How It Works · Data refreshed daily, snapshot 2026-10-20.