eval — A repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.
sandbox — An isolated environment where AI-generated code or agent actions run without being able to touch anything real.
open weights — A model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.
Why it matters
Model selection is a standing decision for anyone building on LLMs, and this maps who is actually shipping open weightsA model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.Full definition → right now and how the balance against frontier models is moving.
articleLatest Open Artifacts (#23): Laguna S2.1, Inkling, & Kimi K3 Show the Utility of Open Models on the Pareto FrontierFlorian Brand
articleLatest open artifacts (#22): Zyphra, Cohere, and Poolside are expanding the breadth of the ecosystemFlorian Brand
videoLocal Models: Trust, Control, Optimization — Carter Abdallah, NVIDIAAI Engineer