Help us train open Australian models

Compute partners. GPU time, hardware or cloud credits. Available compute limits the size of what we can train. We acknowledge partners in the model cards and technical reports they helped make possible.

Research collaborators. Universities, researchers and organisations working on Australian English, Australian knowledge or evaluation data. Better Australian benchmarks help everyone building models here.

Early testers. Organisations that want to try our models on real Australian workloads and tell us where they fall short. We use that feedback in the next training run.