Which AI model to use for each task

There are more than 300 AI models on OpenRouter alone, a service that brings together models from dozens of companies. You don’t need to know them all. You need to know which one is good for what you’re doing.

That’s why we built the AI Ranking.

Why “the best model” doesn’t exist

A model that’s excellent at coding can be expensive and slow for summarising 500 emails. A small, cheap model can be perfect for sorting invoices but weak at writing something with feeling.

So the ranking is split into ten categories, from Coding to Summaries, including Translation and Fast & Cheap.

How to read the ranking

• The score (0 to 10) says how well the model serves that task.
• The price is in dollars per million tokens. A million tokens is roughly 700,000 words. For personal use, it’s almost always pennies.
• “Free” means there’s a version you can use without paying.
• Provisional podium means the results still come from the first evaluation pass.

Choose in 2 minutes

1. Open the ranking and pick the category for your task.
2. If you don’t want to pay, click Free only.
3. Try the top model with a real task of yours.
4. If you don’t like it, move to the second.

The ranking is an AI model’s opinion of other models, based on their published descriptions. It’s a great starting point, but the final test is always your own work.

What if I just want one app?

For most people, one of the big apps (ChatGPT, Claude or Gemini) covers everything day to day. The ranking helps when you want the best at a specific task, or when you want to save money.

See also the guide to using AI for free.