alphaXiv Launches Autoresearch to Turn Static arXiv Papers Into Live AI Experiments

A new platform on OpenResearch allows Claude and Codex agents to replicate and experiment on academic papers from the arXiv repository. Autoresearch uses these AI agents to automate the verification and expansion of research findings.

Tinker API enables agents to launch concurrent reinforcement learning runs during post-training. The workflow transforms academic papers from static artifacts into live research trees that branch as agents conduct their own experiments.