Turing

Turing: an agentic self-improving harness from first principles that provides maximum observability and customization. So think of Claude Code or Codex but Turing is truly open and built specifically for AI model research. Turing runs using open source models like Nemotron, so you can always trace any decision made by the agents and find the rationale behind the decision. This is super important when doing research! But what makes it even more important is that Turing is a self-improving harness. So it researches the best ideas, comes up with hypothesis, runs experiments, and makes decisions on which ideas to keep or skip. All of this entirely readable in a very user friendly manner and you can even customize the agentic system design. We created our own AI research harness (a baseline image model trained on our MacBook from scratch) and and Turing was able to make significant improvements on it!