Provenance Record

Well-structured News piece with clear Overview / What the Model Does / What We Know / What We Don't Know / Analysis sections. Technical depth is appropriate for a general-tech audience; the article names specific tasks (laundry, espresso, box assembly, air fryer, gear rotation) and cites numeric success rates verbatim from the primary TechCrunch report. At 918 words it sits comfortably inside the News 400-1200 word range.

All three sources fetched and verified. (1) https://techcrunch.com/2026/04/16/physical-intelligence-a-hot-robotics-startup-says-its-new-robot-brain-can-figure-out-tasks-it-was-never-taught/ — confirmed: April 16 release, 5%→95% air fryer sweet potato success after ~30 min prompt refinement, matching specialist RL models on laundry/espresso/box assembly, Balakrishna gear-set quote verbatim, Levine 'more than linearly' threshold quote verbatim, lack of standardized robotics benchmarks, and the limitation that the model cannot execute complex multi-step tasks from single commands. (2) https://arxiv.org/abs/2604.15483 — confirmed: paper title 'π₀.₇: a Steerable Generalist Robotic Foundation Model with Emergent Capabilities', submitted April 16 2026, abstract describes zero-shot cross-embodiment generalization, diverse context/multimodal prompting, training on multi-robot data plus suboptimal autonomous episodes including failures plus non-robot sources. (3) https://techcrunch.com/2026/03/27/physical-intelligence-is-reportedly-in-talks-to-raise-1-billion-again/ — confirmed: ~$1B round at >$11B valuation (doubling the ~$5.6B from four months earlier), ~$1B raised to date (so post-round ~$2B total), ~80-person two-year-old company, Founders Fund / Lightspeed / Thrive / Lux investor list, and the Lachy Groom 'no limit to how much money we can really put to work' quote verbatim.

Every numeric and quoted claim maps cleanly to the cited source. The 'past the 2 billion dollar threshold' phrasing is a fair arithmetic summary of the source's '$1B raised to date + $1B new round' framing. The attribution of compositional-generalization framing to the company and the zero-shot cross-embodiment claim to the paper are both accurately sourced. No hallucinations detected.

High-quality News submission. Cryptographic integrity intact, all three sources independently verified against inline claims, neutral tone with explicit caveats about benchmark limitations and out-of-the-box performance. Ready for publication.