ARC Prize Foundation AI Releases: ARC-AGI-3 & ARC-AGI 2

arcprize.org ↗

ThursdAI — the weekly AI news podcast hosted by Alex Volkov — has covered 2 ARC Prize Foundation releases since Mar 2025, most recently ARC-AGI-3 on Mar 26, 2026. Highlights include ARC-AGI 2, ARC-AGI-3. Every entry below has primary-source links, key numbers and the episode segment where we covered it live.

2 releases2 episodesMar 2025 – Mar 2026

March 2026 1

ARC Prize Foundation
Benchmarks & Evals

ARC-AGI-3

ARC-AGI-3 launches: humans score 100%, frontier models under 1%

ARC Prize launched ARC-AGI-3, an interactive agentic reasoning benchmark of turn-based puzzle games designed to test human-like generalization in novel abstract environments. Humans hit a 100% pass rate while top frontier models score under 1%, which the panel welcomed as a healthy reality check against AGI-is-here rhetoric and easy score inflation.

<1% ARC-AGI-3 frontier model scores100% Human completion on ARC-AGI-3
ARC Prize announcement (X) ↗ARC Prize site ↗
🎙️ Hear our coverage →
#benchmarks#reasoning#agents

March 2025 1

ARC Prize Foundation
Benchmarks & Evals

ARC-AGI 2

ARC-AGI 2 benchmark revealed, thinking models score just 4%

The ARC Prize Foundation revealed ARC-AGI 2, the next iteration of the abstract reasoning benchmark. Base LLMs score 0% and even thinking models only reach about 4%, showing how far current frontier models remain from human-level fluid intelligence.

0% base LLM score on ARC-AGI 24% thinking model score on ARC-AGI 2
X announcement ↗
🎙️ Hear our coverage →
#benchmarks#reasoning