AI SEO Benchmark
Digital marketing is being impacted by AI in everyday workflows. But is AI good enough to be trusted for your business? We've designed a unique benchmark for the SEO industry to do just that.
The LLM Leaderboard
Updated Sep 10, 2025
| Language Model | Score (out of 100) | % Diff vs Prev Model | Release Date |
|---|---|---|---|
| Claude Opus 4.1 | 84% | 0% | Aug 4, 2025 |
| Chat GPT-5 | 83% | +3% | Aug 7, 2025 |
| Copilot (GPT-5) | 83% | +6% | Aug 4, 2025 |
| Gemini 2.5 Pro | 82% | 0% | Mar 25, 2025 |
| Perplexity (pplx-7b-online) | 78% | -5% | Sept 8, 2025 |
| Claude Sonnet 4 | 77% | -5% | May 22, 2025 |
| Llama 4 | 75% | n/a | Apr 5, 2025 |
| Deepseek V3.1 | 77% | -2% | Aug 20, 2025 |
| Grok 4 | 77% | +6% | Jul 09, 2025 |
| Gemini 2.5 Flash | 69% | -5% | Mar 25, 2025 |
How We Test AI SEO Performance
- Expert-level SEO questions, updated regularly, focused on in-the-field experience to measure how useful LLMs are at understanding SEO problems.
- The questions are multiple choice and verified by a panel of SEO professionals.
- Based on our test of a sampling of questions with SEO professionals, an expert-level search marketer would be expected to achieve a rating of 89%+ considering that SEO professionals typically specialize in one field.
- SAMPLE QUESTION: Instead of manually penalizing sites that are breaking Google's linking policies, Google has started: (select all that apply)
- Deindexing sites
- Not passing their authority to linked sites
- Quietly dropping their overall ranking
- Changing the SERP to show more ads
- Answer - B & C
Have questions on what this data means to your business? Check out our blog where we will regularly post updates about the progress of LLMs and how it's impacting digital marketing.
