The AI models we test against.
Every audit runs across the AI platforms your customers actually use to find recommendations. The exact engines we test are listed below — when a new model joins the lineup, this page updates with it.
6 platforms currently running 10 different models. Updated automatically.
We don't keep this list a secret. These are the AI tools your potential customers ask for recommendations, and these are the ones we test you against.
Makers of Claude. Widely used in business and professional workflows where reasoning matters.
- ↳ claude-opus-4.8
- ↳ claude-sonnet-4.6
An open-weights challenger from a Chinese lab. Cost-efficient and gaining adoption in technical communities.
- ↳ deepseek-v4-flash
Makers of Gemini. Native AI integrated into Android, Workspace, and the Google search experience itself.
- ↳ gemini-2.5-pro
- ↳ gemini-3.5-flash
Makers of ChatGPT. The market leader. Hundreds of millions of consumers ask it for recommendations every day.
- ↳ gpt-5-mini
- ↳ gpt-5.5
An AI-powered search engine built around recommendation queries. High purchase intent.
- ↳ sonar
- ↳ sonar-pro
Makers of Grok, built into X (formerly Twitter). Growing fast on current-events queries.
- ↳ grok-4.3
We don't rely on one lucky answer. Every audit hits hundreds of AI recommendation queries across the live lineup, extracts the patterns AI keeps mentioning, and turns every gap into a fix on your action tracker. As new models join the lineup above, this number rises with them.
Questions about the lineup, answered.
Which AI models does AI Rank test against?
Six platforms across multiple model versions: OpenAI (ChatGPT), Anthropic (Claude), Google (Gemini), Perplexity, xAI (Grok) and DeepSeek. The full lineup is listed above and updates automatically as models are added, promoted or retired.
How often does the AI model lineup change?
As often as the AI industry ships. We monitor new model releases daily, get alerted internally when a new model from a tracked platform appears, and add it to the testing lineup once we've reviewed it. This page reflects the live state of our registry, so when you see a model listed it is genuinely being tested in audits.
Why test multiple models per AI platform?
Each model gives different answers. The consumer-facing model a free user hits and the paid flagship a business user pays for don't recommend the same businesses. We test the consumer tier AND the paid flagship for every platform so your visibility score reflects what real customers across all usage tiers actually see when they ask AI for a recommendation.
What happens when a brand-new AI model is released?
It goes into our preview tier first, where we observe its behaviour and cost before fully promoting it. Preview models run on a sampled basis so observation cost stays bounded. Once we're confident the model is stable and our customers genuinely encounter it in the wild, it joins the main rotation and contributes to every audit going forward.
Do all plans get tested against every AI model?
Yes for the established models. Preview-tier models — new arrivals we're still observing — are sampled to keep observation cost bounded: one prompt per audit on Starter plans, half the prompts on Growth and Pro plans. Once a preview model graduates to the live lineup, every plan gets full coverage on it.
Why isn't every AI tool on the testing list?
We test the platforms your customers actually use to find local businesses. There are hundreds of AI models in the wild; we pick the ones with meaningful consumer adoption in the UK and skip the rest. As any platform gains real-world traction with the kind of people who hire local businesses, we add it to the lineup.
How is this different from just asking ChatGPT to check my visibility?
One answer is noise. ChatGPT, Claude, Gemini, Perplexity, Grok and DeepSeek all give different recommendations, and the same model returns different answers to the same question on different days. We run up to 540 recommendation tests per audit across the full lineup above, capture each response in full, and extract the reasoning patterns AI keeps repeating. Single answers vary wildly; the patterns we extract across hundreds of tests don't.
See how AI sees your business right now.
30-second free check. No signup. No card.
Run my free check →