传闻 OpenAI 实现递归自我改进

Eco @IndependentEco原推发布于 2026.10.12

在 X 上打开原推

I’m seeing rumors this morning that OpenAI has achieved recursive self improvement. Apparently Sol 6.1 and Astra 6 are performing better than when they released. I have been spending most of my time on the Claude stack. I’m going to pivot back to Codex and see if these rumors are true. Do you see a difference?