Stories
What we build, who we build it with, and what we learn along the way.
Benchmark
1 Million AI Decisions: $16 with Zuse vs. $1,322 with Claude Sonnet
Zuse and Claude Sonnet on the same customer messages: a comparison of routing cost, accuracy and when to send a decision for review.
Model release
Building Zuse, a European decision model
Jev showed that a model which only decides is useful. We built one for Europe: Jev’s requests, hosted in the EU, measured in 27 European languages.
Recognition
Bineric in the Norwegian government’s AI report
A government study of Norway’s AI ecosystem named Bineric as a private company building Norwegian language models.
Customer story
Describing 42,000 works by Edvard Munch
How the Munch Museum’s curators and Bineric built Collective AI: vision models and NorskGPT draft, curators decide.
Looking for the papers? They are on the Research page.