Angela Jiang @angjiang · 29 Jul 2026
GPT-5.6 Sol has been used to solve open problems in mathematics. So why was it struggling with ARC-AGI-3, a benchmark of 2D puzzle games? We investigated. The harness was not letting it remember what it had learned. We found that enabling two API settings tripled our scores https://t.co/0mv05MnSn2
1 228 844Views
9 409Likes
591Reposts
391Replies
221Quotes
0Bookmarks
Is that a lot?
237×vs this author's median5 188 views is typical
83Percentile for this authorof 6 recent posts
424×vs 10K–100K median2 896 views is typical
34.6× audienceReachviews ÷ followers
0.86%Engagement rateof viewers reacted