Model decision tool

Compare models for your use case.

Choose a task, then compare normalized benchmark evidence with dated provider pricing. Missing evidence stays missing.

Step 1

Set the decision

Loading comparison data...

Step 2

Start with four evidence-backed answers

Step 3

Check the comparable evidence

Cost scenario: 1M input + 1M output tokens in USD. Table headers sort with click, Enter, or Space.

? Ranking and value methodology

Performance

Equal-weight mean of each model’s normalized benchmark scores within the selected use case. Offers for the same model share one rank.

Best value

Best value = 2 × performance × cost efficiency ÷ (performance + cost efficiency). Cost efficiency = 100 × the lowest positive scenario cost ÷ the model’s scenario cost.

Evidence limits

Use cases need at least two comparable models. The table retains every priced model plus the 100 highest-scoring unpriced models; prices, dates, and sources are never inferred.