A Tinker research grant.

We are glad to share that Thinking Machines has awarded Kōzu AI a Tinker Research Grant — $5,000 in compute credits on Tinker, their platform for fine-tuning and reinforcement-learning workloads.

It is a small lab's favorite kind of news: not a headline, but runway. Credits are hours, and hours are experiments we can now run instead of queue.

Why it lands where it does

The grant fits our stack almost exactly. Stellarator, our orchestration layer, is already built as a control plane for Tinker — so these credits do not go toward setup or plumbing. They go straight into the loop we already run every day: distill, fine-tune, evaluate, repeat, across the model line from Deimos through Europa to Ganymede.

More runway on that loop means more of the thing we actually care about — tighter reasoning traces, shorter chains of thought, and models that spend their token budget deliberately.

The pursuit

This grant accelerates the advancement of our models in the pursuit of reasoning that is earned, not assumed — intelligence that does more with fewer tokens, and proves it on the tasks that matter.

Our thanks to the team at Thinking Machines. We are excited to see what we can make with the extra time on the clock.