4 ms·
The article doesn't mention one of the most complex benchmarks - ARC challenge. All models suck in it https://arcprize.org/leaderboard https://arcprize.org/lead
by godjan 2y ago
The article doesn't mention one of the most complex benchmarks - ARC challenge. All models suck in it https://arcprize.org/leaderboard https://arcprize.org/leaderboard
but Gemini and Claude still suck much worse then ChatGPT models
- nolist_policy 2y agoThey haven't tested Gemini 2.5 Pro yet.
- usaar333 2y agoThey have. https://x.com/arcprize/status/1905361753678246308 https://x.com/arcprize/status/1905361753678246308