所有国模的宣发都是经过专业训练的，都说自己是第一
ERNIE 5.5 Flash Preview just landed! 🔥
Built on a new architecture designed for efficient inference, the latest open-weight foundation model in the ERNIE family delivers stronger performance in coding, 3D modeling, knowledge work and content creation.
It keeps decoding fast while posting competitive scores on Terminal-Bench, AutomationBench, and other benchmarks.
Here's the rundown:
- 280B total parameters, 14B active per token
- 1M-token context window
- Generation speed above 200 tokens/sec
Model weights will be released by November 25.
159
Actually, the problem with xAI right now isn’t just with Grok Bot. Since Grok 4.5, I’ve been spending an average of $100,000 per month on xAI. But recently, the Grok quota is only 1/4 of what it used to be—cut by 75%. In other words, what used to be a week’s quota has now become a month’s quota, so the cost has multiplied several times.
The following two images show the weekly quota of the Grok subscription I tested previously.
I don’t know if the Grok team understands what this means: it means that what I’m consuming is not top-tier AI, yet the price I’m paying is far higher than Claude.
3
2
755
Previously, xAI’s tokens were very cost-effective. It offered a massive amount of tokens. Although the intelligence it provided wasn’t the smartest, at that price there was really nothing to complain about. But now, its price far exceeds Claude’s, while there is still a big gap in intelligence.
126
XCrun retweeted
为牛市准备好策略工具
为了在排名上超过其他聚合器,部分聚合器的报价总是会虚报价格
对比实际执行，有时会故意降低价格去吃用户的滑点
这对于链上策略是比较难接受的
sextant solver 现已开源
搭配 0x 的 settler 使用
respect @0xProject
github.com/labrinyang/sextan…
这样做的好处是，在不同的书籍里，每个释义 他都有对应的senceID， 那么你在不同的书籍里收藏的单词 属于同一个senceid，就可以归类在一起，跨书籍的收藏 同义 都可以做到归一，个人认为这将是项目迈向标准化必不可少的步骤。
1
116
以前我们的工作模式是每一本书都是新的开始，每次都要重新生成解释，即便这个单词解释过几百遍，在有了我们自己的词典后，大模型只需要做判断题。每本书的词典也将会简化为映射模式，例如单词A的释义A，json里只需要填写AA，就能解决。复用性大大提高
161
