Home › Which AI Quiz
Which AI Is Right for You?
Ten questions, about two minutes, and you get three models ranked against what you actually need. We score every model in our database on your answers rather than handing you the same top three everyone else gets. Nothing is stored and there is no signup.
Sources: Model versions, free tier limits, context windows, and regional availability come from each provider's own documentation, recorded in our model database with a last updated date on every entry. For independent context on how fast capability and cost are shifting, Stanford University's Human-Centered AI Institute publishes the AI Index Report. On how people actually use and feel about these tools, Pew Research Center's artificial intelligence research tracks public attitudes and adoption. Free tiers change without notice, so check the provider's current pricing page before building a workflow around one.
What This Quiz Measures
The quiz sorts you on four things rather than testing knowledge. The first is your primary task, meaning whether you mostly write, code, research, analyse data or work with documents. The second is context length, meaning how long the material you feed a model typically is, since this is the specification that most often decides the answer. The third is your budget position, meaning whether you are strictly on a free tier or can justify a subscription. The fourth is your data sensitivity, which for business users in the region is frequently the binding constraint rather than capability.
No answer is wrong and there is no score. The output is a recommendation with reasoning attached, so you can see which of your answers drove it and override it where your circumstances differ.
How Each Result Maps to a Use Case
A writing or long document result points towards models with large context windows and strong instruction following, because the practical bottleneck is how much source material fits in one prompt rather than raw reasoning ability. A coding result points towards models with strong recent benchmark performance on code tasks, with the caveat that these rankings shift with almost every release.
A research result points towards tools that ground answers in sources you supply and return citations, since the failure mode in research is a confident answer you cannot verify. A data or analysis result points towards models that handle structured output reliably, because the value is in getting a table or a schema you can use rather than prose about your data.
A privacy constrained result narrows the field sharply, and often to a different answer than capability alone would suggest. Where conversation data is stored, and under which jurisdiction, matters more than a few points of benchmark difference for anyone handling client information.
How to Interpret Your Result
Treat the recommendation as a shortlist of one or two rather than a verdict. Free tiers change often enough that the right answer this quarter may not be the right answer next quarter, so the reasoning behind the result is more durable than the name it gives you. Read the linked model page before committing, particularly the section on what the free tier actually includes.
Where two models score closely, test both on one real task rather than deliberating. Ten minutes with your own material is more informative than any comparison table, including ours, because the differences that matter at frontier level are usually about how a model handles your specific kind of input.
Retake the quiz when your work changes rather than when a new model launches. The questions are about your requirements, and those move more slowly than the model landscape does.