2 ms·
The nonprofit Arc Prize says that Grok achieves a new state-of-the-art score on its ARC-AGI-2 test — another difficult benchmark that consists of puzzle-like pr
by andsoitis 1y ago
The nonprofit Arc Prize says that Grok achieves a new state-of-the-art score on its ARC-AGI-2 test — another difficult benchmark that consists of puzzle-like problems where an AI has to identify visual patterns — scoring 16.2%. That’s nearly twice the score of the next best commercial AI model, Claude Opus 4.