Mark Griffiths (@codergriff) 在 Piclur：1) incredible cost/efficiency progress, 2) crazy that this level of intelligence was almost there 12 months ago, 3) what level of intelligence do labs have right now that's too expensive to release? / Piclur

Mark Griffiths

年12月12日

1) incredible cost/efficiency progress, 2) crazy that this level of intelligence was almost there 12 months ago, 3) what level of intelligence do labs have right now that's too expensive to release?

A year ago, we verified a preview of an unreleased version of @OpenAI o3 (High) that scored 88% on ARC-AGI-1 at est. $4.5k/task Today, we’ve verified a new GPT-5.2 Pro (X-High) SOTA score of 90.5% at $11.64/task This represents a ~390X efficiency improvement in one year

arcprize's tweet image. A year ago, we verified a preview of an unreleased version of @OpenAI o3 (High) that scored 88% on ARC-AGI-1 at est. $4.5k/task

Today, we’ve verified a new GPT-5.2 Pro (X-High) SOTA score of 90.5% at $11.64/task

This represents a ~390X efficiency improvement in one year

United States 趋势

1. The AsterDEX 33.2K posts
2. Pro Bowl 16.3K posts
3. Happy Festivus 2,753 posts
4. Christmas Eve Eve 60.4K posts
5. The AAVE 199K posts
6. FINALLY DID IT 551K posts
7. Cam Jurgens N/A
8. #AvengersDoomsday 114K posts
9. The PENGU 220K posts
10. Jordan Davis N/A
11. Derrick Brown N/A
12. Larry Nassar 14.7K posts
13. The WET 23.5K posts
14. Mainz Biomed N.V. N/A
15. Nashon Wright N/A
16. Jalen Carter N/A
17. Carry Water 1,292 posts
18. Joe Alt N/A
19. Tennis 22.1K posts
20. #tuesdayvibe 2,734 posts

Something went wrong.

Mark Griffiths

ARC Prize

United States 趋势