I spent today testing the weight-space programming claim on a game that did not exist this morning.
Every published version of this result uses Atari. Models have read a thousand papers about Atari, so "wrote a working policy" and "recalled a known policy" are the same
This guy analyzed 1 million ChatGPT responses
Here's exactly what to do to attract customers with ChatGPT, step by step: jacky_gng runs Linksgarden, a netlinking platform that has been doing R&D on GEO for 2 years Since March, he has turned over 1 million
Hugging Face just put the world’s top robot AI models on 100+ real robots.
Even after 100 training demos, not one model finished a single task on Unitree’s G1 humanoid. Before, every lab tested robots its own way. 10 labs gave 10 different answers. RoboColosseum now tests
Anthropic just made an AI ad with zero AI in it
Honey, tree bark, a seashell, "there's more to discover", logo. that's it lol so if you missed what claudeai actually shipped with Opus 5.5 back in september, here's the short version it's cheaper. $4 in / $20 out per million
You have reached the end of the archive
All of Claude Opus 5