Big Walk is being review bombed after House House revealed AI was used on “well under 0.1%” of its code
The studio says a chatbot helped fix a Unity bug, with the final solution using around 10 lines of heavily modified code
Elo 涨了 8 分,但模型大概率一行代码都没改。
权重是上周才开放的,榜单上对战的还是同一个 Inkling 2 Small。竞技场模式的 Elo 本来就会随着对战样本增加、对手池更新而持续浮动,8 分在这个量级里基本属于噪声,拿它讲发布后势头更好没什么意义。
我反而会盯着它对 Kimi K3 的 32 分领先看。按标准 Elo 公式换算,32 分大约对应 55% 的预期胜率,差不多每九把赢五把,榜首和次席的差距远比名次看起来薄。一个 Small 尺寸模型在 HTML 这种垂直任务上,用对手十五分之一的输出价格拿到开放权重第一,这件事比 Elo 涨跌更值得琢磨。
https://t.co/b7FlkwwiaI
UPDATE: Inkling 2 Small by @thinkymachines is now open-weight and is ranked #1 among open-weight models on Overall HTML with an Elo of 1384, up 8 Elo points since its initial release last week.
It currently holds a 32-point lead on Kimi K3 by @kimi_moonshot and is still the only American model to be #1 among open-weight models on DesignArena.
Congrats to the @thinkymachines team on this achievement!