Research
October 2 2026
A Pareto frontier Database Reasoning model in 4B parameters
Our first submission, Native mini, with only 4B parameters, scored 76.86% on BIRD's hidden test set.
February 18 2026
Native ranks #1 on Spider 2.0
Native ranks #1 on Spider 2.0 with 96.53% accuracy, the first result to break the 95% barrier.