SearchAgent - PopQA: leaderboard
SearchAgent PopQA slice measures exact match for search agents on long-tail factual QA.
Metric: Exact Match (%). Source: huggingface.co. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | ReSeek-Qwen2.5-7b-Instruct | 50.1 | |
| 2 | ZeroSearch-Qwen2.5-7b-Instruct | 48.8 | |
| 3 | ZeroSearch-Qwen2.5-3b-Instruct | 44.8 | |
| 4 | ReSeek-Qwen2.5-3b-Instruct | 43.4 | |
| 5 | Search-R1-Qwen2.5-7b-Instruct | 39.7 | |
| 6 | RAG-Qwen2.5-7b-Instruct | 39.2 | |
| 7 | Search-R1-Qwen2.5-3b-Instruct | 37.8 | |
| 8 | R1-Qwen2.5-7b-Instruct | 19.9 | |
| 9 | Direct-Inference-Qwen2.5-7b-Instruct | 14 | |
| 10 | Search-o1-Qwen2.5-7b-Instruct | 13.1 |
Interactive version: theaggregate.ai/benchmark?slug=searchagent-popqa · How It Works · Data refreshed daily, snapshot 2026-10-20.