GPT-6 Astra represents a step-function change in model capability for interactive reasoning problems. It scores 66% on ARC-AGI-3 using our standard harness, and nearly 100% with a continuous conversation harness and custom compaction, at a cost of roughly $360 per game.
In fact, the continuous harness version significantly outperforms our human baseline in action efficiency across almost all levels. When we examined the reasoning chains to understand how the model operates, we found it performing highly efficient, on-the-fly symbolic world modeling for each game and level. It goes as far as developing its own shorthand DSL to represent in-game situations -- essentially a game-specific algebraic notation.
Overall, Astra exhibits symbolic modeling behaviors we had previously only seen with sophisticated harnesses -- so harness capabilities are increasingly shifting into the model itself.
We see Astra as a major breakthrough in model intelligence.
Read our post on Astra and what these results mean: https://t.co/wJnYxEqYNI
A BBC interviewer asked a Nobel physicist the simplest question in the world: why do two magnets push each other apart. He refused to answer it. What he did instead is the best seven minutes on thinking ever filmed.
The man is Richard Feynman, at home in 1983, no blackboard, no notes, sitting in an armchair.
He does not dodge because he cannot answer. He dodges because he wants you to see what the question is actually asking. So he takes an ordinary sentence, that a woman slipped on the ice and broke her hip, and starts asking why, and every answer opens another why, until the ground gives way.
Then he turns it around on the interviewer with a question about ice that sounds trivial and is not.
The point he arrives at is the uncomfortable one. Every explanation is only a translation into something you already accept without explanation. Magnets have no such translation. There is nothing more familiar than magnetism to reduce magnetism to, which is why every answer you were given at school was a picture, not a reason.
It matters more now than it did in 1983. Any tool will hand you a fluent answer in a second. Knowing which questions actually have a bottom is the part that is still yours.
One question. Seven minutes. It is in the video.
Never has a dumber group of people been put in charge of the world than this one. The Trump administration is so moronic, you simply can't parody it. This is the closest we've come to the movie 'Idiocracy' becoming a documentary.
@Rainmaker1973 The invention of agriculture and the following end of our travelling lifestyle. It made owning land and a large number of objects feasible. Inheritance and greed followed soon…
Con le famiglie ancora alla ricerca disperata dei figli, sui media e sui social si sentono e leggono commenti intrisi di superficialità, cinismo e cattiveria su come hanno o non hanno reagito, in quei momenti, quei poveri ragazzi.
Come fosse colpa loro… Gli unici responsabili, gli unici ad avere la coscienza sporca, gli unici delinquenti sono coloro che non hanno garantito la sicurezza in quel seminterrato, coloro che non hanno controllato, coloro che hanno autorizzato, coloro che per soldi hanno ammassato centinaia di ragazzi in un sotterraneo, con fuochi accesi e sistema di emergenza fallimentari.
Nella civile Svizzera, le porte della galera dovranno aprirsi per parecchie persone. Una preghiera per quelle povere anime volate in cielo troppo presto e per i loro cari 🙏🏻