researchaudio

Steinberger chose speed over independence. The agent era accelerates.

Sandboxes turn “the model can do things” into “the model can do things safely

TinyLoRA suggests RL can steer big models with updates small enough to fit in a tweet.

Zhipu's GLM-5 tops open-source benchmarks with a novel async RL framework called SLIME.

What changed in the API, what failed during agent testing, and how to design tool-using systems that stay inside the rails.