返回
RCreddit.com
15
·20小时前·开发者社区 · RSS

The overlooked GPT-5.6 metric: OpenAI’s internal AI usage jumped 22x in six months

查看原文
OpenAI订阅权益

热度趋势

新上榜
最近 24 小时与此前 24 小时对比 · 7 天曲线

百分比基于当前可用热度信号,而非评论数或独立用户人数。

推荐理由

这条记录涉及编程工具或代码能力更新,适合开发者评估工作流变化和可复用价值。

Everyone is focused on the benchmark scores, but one number in OpenAI’s GPT-5.6 release caught my attention much more. OpenAI says its internal agentic token usage increased roughly 22x over the last six months, while the share of research compute going toward internal coding inference grew 100x. That feels more significant than another model gaining a few points on a benchmark. The company building frontier models is increasingly using those same models inside the process of building the next generation. I don’t think this means “recursive self-improvement” in the dramatic sci-fi sense. Humans are still setting the objectives, running the infrastructure and deciding what gets trained. But it does look like the R&D loop is starting to compress: better models help researchers work faster, which helps produce better models, which can then accelerate the next cycle. If that trend continues, the important question may stop being how much better is the next model? and become how much faster can an AI lab improve once AI is deeply embedded in its own research process? Curious where people here draw the line between ordinary AI-assisted engineering and something qualitatively different.

The overlooked GPT-5.6 metric: OpenAI’s internal AI usage jumped 22x in six months · BuzzRadr