Newsletter September 10, 2026
How this comes together. This page is automatically populated daily from public sources (X, Hacker News, GitHub, YouTube). The headlines come from the source, the Dutch sentence below is machine‑generated. Click through to the source before basing anything on it.
Repos, releases & updates
- I tested 6 LLMs in 6 days on YouTube. Crazy pace of new model releases, almost every single DAY. I hope they stop for a while and get me som
The article describes a test of six LLMs in six days, showing that new models appear almost daily and that indie developers need to quickly evaluate which they want to use. - Meta FAIR Releases AI Research Preference Models (RPMs): Frozen LLM Judges That Decide Which ML Experiment Gets the GPU, Lifting AIRS-Bench
Meta FAIR introduces AI Research Preference Models that act as fixed LLM judges to decide which experiments get GPU access, which can help indie developers plan their compute budget. - GPT-6 Astra Is Here. But Can We Trust the Leaderboards? @OpenAI released GPT-6 Astra, calling it its most capable model yet across coding,
OpenAI has launched GPT‑6 Astra, a model that, according to the announcement, is strong in code generation, and the coverage questions whether the leaderboards are reliable, which is relevant for indie developers who want to assess performance.