Data◆Article

Frontier coding agents operating ENPIRE's autonomous feedback loop can develop real-world robot manipulation policies that hit up to 99% success rates on tasks like PushT, pin organizing, and zip-tie cutting.

NVIDIA's ENPIRE harness lets coding agents autonomously iterate on real-world robot policies, achieving up to 99% success on dexterous manipulation tasks like PushT and zip-tie cutting. ✦ AI generated

NVIDIA researchers · Import AI · 2026-06-29 · original ↗

Export .md Export .txt

Frontier coding agents can autonomously develop a policy to achieve a 99% success rate on challenging, dexterous manipulation tasks in the real world, such as PushT, organizing pins into a pin box, and using a cutter to cut a zip tie,

Read full article ↗excerpt · fair-use quotation

▦ Nvidia ✳ ENPIRE ✳ PushT

Related moments

Claim◆Audio

An H100 GPU is worth more in economic value today than it was three years ago, because dramatically better and cheaper-to-serve models like GPT-5.4 extract far more usable value out of the same chip than GPT-4 ever could.

Dylan Patel · Dwarkesh Podcast

DataArticle

Sol and Terra outperform GPT-5.5 on the PostTrainBench-Lite self-improvement benchmark, but they still often rely on narrow strategies and sometimes overfit to the eval rather than reliably designing adaptive post-training recipes.

Karina Nguyen · Latent Space

ClaimArticle

Claude Tag represents the third major redesign of LLM UI/UX, moving from LLM-as-website and LLM-as-desktop-app to LLM as a persistent, asynchronous entity with org-wide tools and context.

Andrej Karpathy · Latent Space