Hackernews posts about ARC-AGI-1

1. 01
   

   newArc AGI 1 400/400 No LLM running on my MacBook

   ▲ 25CEOs29ddiscusssavesaved
2. 02
   

   newDeepSeek v4.1 Flash 1017 Scores on ARC-AGI-1/2arcprize.org

   ▲ 1matt115ddiscusssavesaved
3. 03
   

   newSamsung released a 7M model that achieved 45% on ARC-AGI-1arxiv.org

   ▲ 34chintler394d💬 12savesaved
4. 04
   

   newF. Chollet: We were able to reproduce the findings of the HRM paper on ARC-AGI-1twitter.com

   ▲ 5Anon84448ddiscusssavesaved
5. 05
   

   newInkling is now the highest-scoring open-weight model on both ARC-AGI 1 and 2twitter.com

   ▲ 3thebricklayr112d💬 1savesaved
6. 06
   

   new150M-parameter reasoning model sets new cost-accuracy frontier on ARC-AGI-1huggingface.co

   ▲ 3meander_water87ddiscusssavesaved
7. 07
   

   newOver the course of 19 months, the cost to reach 75% on ARC-AGI-1 fell 99.95%twitter.com

   ▲ 2theanonymousone35ddiscusssavesaved
8. 08
   

   new44% on ARC-AGI-1 in 67 centsgithub.com

   ▲ 1evilmathkid246d💬 1savesaved
9. 09
   

   newARC-AGI-1 and 2 LEADERBOARDarcprize.org

   ▲ 1doener267d💬 1savesaved
10. 10
    

    newTranscendPlexity: 540/540 ARC-AGI-1/2/3, 13 tasks with 0% AI solve rate, solvedgithub.com

    ▲ 1wormsWorld168ddiscusssavesaved
11. 11
    

    newPoetiq announces new SOTA on the ARC-AGI-1 and 2poetiq.ai

    ▲ 1fofoz344ddiscusssavesaved

· edit