An AI agent screened over two million crystals, invented four new superconductors, and a lab confirmed all four are real -- one designed from scratch. Then two papers finally pin down the exact math of training a reasoning model, and why nearly half your training data may teach it nothing. Plus: giving an agent a memory lets a tiny model act like a giant -- and teaches it to trust its own stale notes over the facts. And in the news, a day defined by closed boxes: a top coding model clipping its own reasoning at exactly 516 tokens, newest models inventing tool fields that don't exist, and a Claude Code session leaking a stranger's live credentials.

Chapters

0:00 The Model That Cuts Off Its Own Thoughts
1:19 The Headlines
5:27 Intro
6:57 An Agent Found Four New Superconductors
12:23 The Exact Math of Teaching a Model to Reason
18:42 Agents Learn to Remember -- and to Suck Up to Themselves
24:36 Wrap-up

Links

The Model That Cuts Off Its Own Thoughts -- https://github.com/openai/codex/issues/30364
An Agent Found Four New Superconductors -- https://arxiv.org/abs/2604.23758
The Exact Math of Teaching a Model to Reason -- https://arxiv.org/abs/2607.00152
Agents Learn to Remember -- and to Suck Up to Themselves -- https://arxiv.org/abs/2606.29961