SearchGEO: leaderboard
Large language model (LLM)-based search agents synthesize open-web content into actionable recommendations on behalf of users, creating a risk that attacker-published pages are transformed into endorsed claims. We introduce SearchGEO, a controlled evaluation framework for measuring endorsement corruption in LLM-based web-search agents, combining a web-evidence manipulation pipeline, a five-mode at
Metric: ASR↓ (%) (self-reported). Source: benchmarklist.com. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.6 | 0 | #90 |
| 2 | GPT-5.4 Mini | 0.8 | #212 |
| 3 | GPT-5.4 Nano | 6.4 | #328 |
| 4 | Kimi K2.6 | 7.2 | #104 |
| 5 | Grok 4.3 | 7.6 | #188 |
| 6 | DeepSeek V4 Pro | 9.1 | #102 |
| 7 | Claude Haiku 4.5 | 12.5 | #280 |
| 8 | Qwen 3.6 Plus | 14.4 | #120 |
| 9 | MiniMax-M2.7 | 17.4 | #258 |
| 10 | DeepSeek V4 Flash | 20.1 | #137 |
| 11 | Gemini 3.1 Pro (Preview) | 21.2 | #58 |
| 12 | Gemini 3.5 Flash | 25.8 | #60 |
| 13 | Gemini 3 Flash | 31.4 | #99 |
Interactive version: theaggregate.ai/benchmark?slug=searchgeo · How It Works · Data refreshed daily, snapshot 2026-10-20.