Arena compares frontier AI models through leaderboards covering text, image, vision, and other areas.
Tracking AI
Tracking AI compares IQ scores assigned to frontier AI models and lets visitors explore the listed scores.

About Tracking AI
Tracking AI is a web tool that shows IQ scores assigned to frontier AI models and lets visitors explore the listed results. It is run by Maxim Lott, who is the executive producer of Stossel TV. The site reports each model's reasoning score, average answer time, and cost per answer, and it includes a view of top AI models over time. Visitors can also see how each AI answered every question.
The site describes two IQ tests. The "Offline" quiz is described as a test made by a Mensa member that has never been on the public internet and is not in any AI training data. The Mensa Norway test is described as a public online IQ test. For IQ tests, questions are presented in verbal form, and vision models are asked the test image. If a model refuses to answer, the site asks the same question up to 10 times. If it still refuses, the site uses the most recent answer it gave and marks that the model refused that day.
The site states that AI models change their answers constantly, so scores may change over time. It also says that it does not expect all major AIs to always answer every question, and that a metric showing which AIs were unwilling to answer certain questions is planned for the future.
Key features
- Lists IQ reasoning scores for frontier AI models
- Shows average answer time and cost per answer for each model
- Displays how each AI answered every question
- Filters results by last day, last week, or last month
- Offers a chart of top AI models over time
- Includes an offline IQ quiz described as never published online
- Shows per-question answers as well as overall scores
- Explains how refusals are handled in scoring
- Uses an offline test that the site describes as absent from public internet and AI training data
- The site states that AI models change their answers constantly, so scores can change
- AIs that refuse to answer may receive scores based on an earlier answer, as described in the scoring method
- A metric for questions an AI was unwilling to answer is planned but not yet available
Tracking AI at a glance
- Website
- trackingai.org/home
- Category
- AI & Machine Learning › AI Data & Analytics
- Type
- Online tool
- Pricing
- Not confirmed. Check the site
- Platforms
- Web
- Open source
- Not stated
- Language
- English
- Good for
- Creators
- Last checked
- Oct 6, 2026 · Online when checked
- Listed
- Oct 5, 2026
Tracking AI alternatives
Tools and sites you could use instead, ranked by how closely they match.
The Hugging Face Open LLM Leaderboard compares language models across evaluation tests including IFEval, MATH, and MMLU-Pro.
Explore leaderboards with expert-driven LLM benchmarks and updated AI model rankings across coding, reasoning and more.
SimpleBench evaluates everyday reasoning by comparing language models with human performance.
LLM Stats compares language models by rankings for intelligence, speed, and price.
Deploybase compares GPU cloud and language model API prices, performance benchmarks, and provider market data.
Popular in AI Data & Analytics
Well-known listings from the same topic.
Tracking AI FAQ
How are AI models scored when they refuse to answer a question?
How are the IQ tests presented to the AI models?
Where can I access the AI models the site tracks?
Tell us and an editor will check it.
Claim the listing and suggest updates yourself. An editor checks every claim first.












