The winner is the strongest practical weekly result
We summarise the models that performed best during the period, but important choices should still be checked against the current daily benchmark.
Reports
AI benchmark reports from hvilkenAI
We publish a weekly summary of AI performance on Norwegian, Swedish and Danish. The latest reports are available with a 2-week delay — contact us for early access.
hvilkenAI evaluates over 350 AI models independently. We have no commercial ties to the AI providers we assess.
Monthly reports
Weekly reports
Earlier weekly reports have been withdrawn.
They were built on measurements we now know were wrong. We are not leaving them up with retroactive corrections, because they cannot be honestly reconstructed.
Price and stability matter
The most expensive model is not automatically best for Scandinavian-language work. The reports show when lower-cost alternatives are strong enough.
Reports are delayed intentionally
Public reports are published with a delay so the most valuable ongoing analysis can remain part of paid access.
Why this is useful
The AI market changes quickly. Providers update models, prices move and performance on smaller languages can vary. Weekly reports make it easier to spot trends over time without interpreting daily benchmark tables manually.
See also