Announcing ARC-AGI-3
The only unsaturated agentic intelligence benchmark in the world
Humans score 100%, AI <1%
This human-AI gap demonstrates we do not yet have AGI
Most benchmarks test what models already know, ARC-AGI-3 tests how they learn
Today we're launching the ARC-AGI-3 Toolkit
Your agents can now interact with environments at 2,000 FPS, locally.
We're open sourcing the environment engine, 3 human-verified games (AI scores <5%), and human baseline scores.
ARC-AGI-3 launches March 25, 2026.
@karooolis Right. And if people like that, cool. I just wish decisions like this had more of philosophical underpinning, instead of someone smashing the "I don't like this, make it illegal" button every few weeks
@karooolis Intervention may have benefits here, but this is fundamentally a Nanny State mindset. This logic can be applied to justify regulation on any scarce good, like Pokemon card ("Fans must have the chance to buy 1st edition Charizard at a fair price...")
@Keivansamani it must be super uncanny hearing two languages simultaneously, having to fiddle with the volume, trying to have a group conversation, etc...