Model guide
The best AI model for each kind of work
Rankings come from LMArena, where people compare two anonymous models on the same prompt and vote for the better answer. Every model listed is live on Rapid Router today, behind one API key, with the price you would pay.
Ranked on LMArena's coding board, where developers compare two anonymous models on the same programming prompt and vote for the better answer. A higher rating means the model wins more of those head-to-heads.
Ranked on LMArena's WebDev board, where each model builds a working web page from the same prompt and people vote on the result. It rewards models that produce code that runs and looks right, not just code that reads well.
Ranked on LMArena's hard prompts board: the subset of real user prompts judged most demanding, with several constraints, specialist knowledge or long chains of logic.
Ranked on LMArena's math board, built from the maths questions people actually ask, from school algebra to proofs. Benchmark scores such as MATH Level 5 and AIME appear on each model's page.
Ranked on LMArena's creative writing board: stories, emails, scripts and copy, judged by people choosing the answer they would rather read.
Ranked on LMArena's instruction following board: prompts with explicit formats, lengths and rules, where the better model is the one that obeys all of them.
Ranked on LMArena's vision board, where people upload an image, ask a question about it and vote between two models' answers. Only models that accept image input are listed.
Ranked on LMArena's overall board, the broadest signal there is: millions of votes across every kind of everyday question.
Ranked by context window, the amount of text a model reads in one request, with its overall LMArena rating alongside so a big window is not mistaken for a good model.
The cheapest models that still rate in the top 40% of LMArena's overall board, sorted by price. Good enough for most work at a fraction of a flagship's cost.