The Soyuz program traces its lineage to the first human spaceflight. Part of flying on the Soyuz includes joining that tradition. I wanted to share some of that experience here. Before we leave for Kazakhstan next week, we participated in that tradition and visited Red Square to lay flowers on the memorials of Yuri Gagarin, Sergey Korolev, and other cosmonauts. I’ve read a lot of books on the subject, but it’s hard to imagine what it was like to witness those first space flights. I can only think it was like seeing self-landing rockets, Artemis, and a moonbase become a reality. These next few years will be amazing to witness!

Interesting paper on prompt optimization. They claim that a single-lineage prompt optimizer just matched GEPA on a smaller rollout budget. Prompt optimization has been drifting toward heavier machinery, with candidate pools, reflection trees, and Pareto-based selection. NPO keeps one lineage. At each iteration it runs the student on the current prompt, collects rollout traces and rewards, and hands a sliding window of recent iterations to a teacher model that rewrites the prompt. There is no candidate population and no search tree. On the two instruction-following benchmarks it spends 3,500 and 6,800 rollouts against GEPA's 3,593 and 6,871, and it stays broadly comparable across 22 TextArena games. The interaction with teacher strength is what makes this interesting. NPO's advantage grows as the teacher model gets stronger, which suggests optimizer-side search complexity has been compensating for weak teacher reasoning all along. Paper: Chat with Paper:

@parshurRam @elonmusk Sure! Based on your Vedic Brahmin lineage, devotion to Shri Ram, and vaidya traditions, here's an HD imagined portrait of you as a 22-year-old scholarly Konkanastha Brahmin male.

Tencent HY proposes environment evolution, which builds off-policy lineages of increasingly difficult verified terminal tasks, keeping long-horizon RL signals alive as Qwen3.6 agents improve. Environment Evolution for Terminal Agents Paper: