OpenAI 在 2 月表示不會公開推出 o3 推理模型後，最近又表示計劃在「數星期內」同時推出 o3 及其下一代 o4-mini，而備受期待的下一代主力模型 GPT-5 就因為開發進度未如理想，需要延遲到「幾個月後」推出。
OpenAI 總裁 Sam Altman 在 X 表
Meta 宣布推出全新一代大型自然語言處理模型 Llama 4，並將推出兩款主要版本：Llama 4 Scout 和 Llama 4 Maverick，並預告即將推出更大規模的 Llama 4 Behemoth 及提高推論效能的 Llama 4 Reasoning。
Llama 4 Scout 配備了 170 億個參數 (109B 總參數
In collaboration with researchers from Tsinghua University, DeepSeek developed a technique that combines methods referred to as generative reward modelling (GRM) and self-principled critique tuning, according to a paper published on Friday. The dual approach aims to enable LLMs to deliver better and faster results to general queries.
ModAI, unveiled on Thursday, is designed to automate content moderation, generate humanlike engagement, and offer customer support. The tool could “slash operational costs by 80 per cent and double
天天有新東西宣佈的 OpenAI本週預告 ，可幫用户上網做研究的深度研究（Deep Research）AI代理功能，很快將推向免費版ChatGPT。
OpenAI是在今年2月初 首先公佈Deep Research 代理人功能給Pro方案用户。Deep Research採用了整合強化學習與多模態推理的方法，能在用户
The big picture: Only days ago we were reporting that ChatGPT's updated image generator was sparking a wave of Studio Ghibli-style recreations across social media. But the wave was just getting started and in mere days it hit the mainstream, becoming a full-blown fad . While CEO Sam Altman is celebrating the newfound attention, the massive growth in GPT's user base, and a fresh round of funding,