← 返回选刊

X · @rauchg / 长帖

Rauch:Fable 5.2 能并行做很多事,但要用上云端沙箱

Vercel 在发布前测过这个模型。他认为这一代模型能承担前所未有的并行工作量,现在的瓶颈是野心和并行算力。

中文改写

Guillermo Rauch 说,Vercel 在 Fable 5.2 发布前就测过这个模型。他的概括很短:astonishingly brilliant。

他看到的一个重点是,这一代模型能承担前所未有的并行工作量。

要把这种能力真正用起来,他认为必须有云端沙箱,没有商量余地。模型能委派和编排的工作,已经超出单台电脑能承受的范围,也可能超出你以为它能承受的范围。

于是现在的瓶颈变成了两件事:一是你的野心,二是你的并行计算能力。

具体做法上,他从 Malte Ubl(@cramforce)那里学到的办法并不复杂:安装 Vercel CLI,配置好插件,然后让智能体使用 /offload-to-sandbox 这个 skill。它会按需要开通和回收云端沙箱。按他的说法,你不必再去做一个所谓的 multi-agent remote harness ralph orchestrator。

这条帖子引用了 Vercel Developers 的公告:Claude Fable 5.2 已在 AI Gateway 上线,模型标识是 anthropic/claude-fable-5.2。公告还写到,它在编程、推理和视觉上比之前的 Fable 更好,并能编排或指导较小的模型处理更难的问题。公告地址是 https://x.com/vercel_dev/status/2119869860096466990 。

如果也想让这一代模型把大量工作并行做完,可以像他写的那样,先从现成的沙箱 skill 用起,而不是先搭一套复杂的远程编排系统。

原文

At Vercel we got to test this model before release. Best way to summarize it: astonishingly brilliant.

At @vercel we got to test this model before release. Best way to summarize it: astonishingly brilliant.

One interesting takeaway: it's very clear that this new generation of models can take on unprecedented bodies of parallel work.

To actually seize this intelligence, you need cloud sandboxes. Full stop. The model can delegate and orchestrate far more than a single computer can handle. Likely far more than you can even imagine it can handle.

Your bottleneck is now ① your ambition and ② your parallel computing power.

Btw, I learned from @cramforce you don't need to overcomplicate it. Install Vercel CLI, setup the Plugin and tell the agent to use /offload-to-sandbox skill. It'll provision and de-allocate cloud sandboxes as needed. You don't need a "multi-agent remote harness ralph orchestrator" 😆