EPISODIO 24

OpenAI o3 VS Gemini 2.5 Pro - Molti Benchmark

• ARC-AGI 1 e 2: https://arcprize.org/blog/analyzing-o3-with-arc-agi; https://arcprize.org/leaderboard
• Virologia: https://www.virologytest.ai/
• PHYBench: [https://phybench-official.github.io/phybench-demo/;](https://phybench-official.github.io/phybench-demo/) https://arxiv.org/pdf/2504.16074
• GeoBench: https://geobench.org/
• VisualPuzzles: https://neulab.github.io/VisualPuzzles/
• NaturalBench: https://linzhiqiu.github.io/papers/naturalbench/

Detti e proverbi

• Le tweet: https://x.com/rose_matt/status/1915355300506325282
• Il nostro benchmark: enkk.me/proverbit

Deep Search Lightweight

• Le Tweet: https://x.com/OpenAI/status/1915505959931437178