OpenAI昨（24）日宣佈推出 以OpenAI o4-mini為基礎的輕量版深度研究（Deep Research）代理人，給ChatGPT所有付費及免費方案用户。
OpenAI表示，由於發現2月推出的Deep Research受到眾多用户歡迎，因此決定提供輕量版本，以擴增現有的使用額度。在用户標
由兩名南韓青年創立的Nari Labs本週二（4/22） 透過GitHub 及Hugging Face，開源了具備16億參數的 文字轉語音模型Dia ，宣稱它不僅與Google NotebookLM播客的品質相當，甚至超越了ElevenLabs Studio與Sesame的開源模型。Dia發布48小時便在GitHub上獲得超過7,800顆
iFlytek on Monday boasted that its Xinghuo X1 reasoning model, a “self-sufficient, controllable” LLM trained with
Why it matters: When Liang Wenfeng launched his advanced AI model DeepSeek on Hugging Face, it marked a turning point for artificial intelligence and the global open-source movement. Its debut shifted the focus from a Chinese national achievement to a broader story about how open collaboration can cross borders and reshape innovation.
MongoDB Developer Relations head and open-source advocate Matt Asay argues that DeepSeek represents more than just Chinese innovation – it shows
OpenAI 近來推出了不少新模型，不過最近 o3 模型在官方和第三方基準測試結果之間的差異，就引發了外界對 OpenAI 透明度和模型測試實踐的爭議。
OpenAI 於去年十二月發布 o3 時，宣稱該模型能夠解答 FrontierMath（一組困難的數學問題）中超過 25
What just happened? Microsoft has introduced BitNet b1.58 2B4T, a new type of large language model engineered for exceptional efficiency. Unlike conventional AI models that rely on 16- or 32-bit floating-point numbers to represent each weight, BitNet uses only three discrete values: -1, 0, or +1. This approach, known as ternary quantization, allows each weight
大型語言模型（LLM）已滲透到各行各業，但享受生成式AI便利的同時，必須瞭解這項新興科技背後的風險，長期耕耘網站安全的OWASP組織已訂定10個主題，想認識AI安全議題可從這裡開始著手。
Microsoft researchers just created BitNet b1.58 2B4T, an open-source 1-bit large language model with two billion parameters and trained on four trillion tokens. But what makes this AI model unique is that it’s lightweight enough to work efficiently on a CPU, with TechCrunch saying an Apple M2 chip can run it. The model is also readily available on Hugging Face , allowing anyone to experiment with it.
Bitnets use 1-bit weights with