11 views · 1 link click
In reply to @andonlabs · Sep 8, 2026We've never seen this before.
The biggest jump in Vending-Bench history. GPT-6 Astra is better at making money and more ethical than Claude Fable 5.1.
Surprising, because:
1. First time ever that OpenAI is #1 on Vending-Bench
2. The best model is no longer the unethical one. t.co/jVM6Fxcblz
@andonlabs If you're still researching Drone-Bench (which I love), I think you'll dig my autonomous research project for UAV-navigation:
alexisrondeau.me/low-light-geol…
Is there a Github repo I can check out and what was your personal biggest surprise or learning so far?
“Not all who wander are lost”Ballpark — “Not all who wander are lost”: Can a UAV learn a city by heart — no GPS, no map on board, just a $4 flight computer?Yes: a from-scratch neural memory that knows Berlin by heart. One downward photo in, (lat, lon, confidence) out — the map is the weights, one city in 3.1 MB. Found by an autonomous loop of coding agents in 81 experiments for $364.↗ alexisrondeau.me