Nvidia DGX Cloud Supercomputer Launches on Oracle Cloud


IBM's CEO questions the return on investment for the huge investment in AI data centers, pointing out that current construction costs are high and the global commitment to computing power is massive, implying the need for a careful assessment of the investment benefits.
Kuaishou's Keling Digital Human 2.0 is now fully launched, enabling users to create expressive digital human videos in just three steps. The new version supports uploading character images, adding voiceovers, and describing performances, generating videos up to 5 minutes long. Compared to the previous version, 2.0 significantly enhances expressiveness with precise control over hand movements and lip-syncing.....
AliQwen APP has launched a learning large model called Qwen3-Learning, trained based on Qwen3, specifically designed for learning scenarios. It provides two free services: photo-based problem solving and homework correction, with no limit on usage. Compared to the paid features of OpenAI and Google, this model offers higher accuracy in recognition, supports multiple languages and cross-cultural problem solving, and integrates resources from over 30 countries worldwide.
Luo Yonghao praised ByteDance's launch of the 'Dou Bao Phone' engineering prototype, believing that AI assistants will be everywhere, completely changing life and becoming an indispensable part. He said that trying to launch an AI phone is a remarkable progress.
The new model matrix codenamed 'Penguin' by OpenAI has been exposed, including the flagship Emperor512, mid-range Rockhopper64, lightweight Macaroni16, and zero-reasoning Mumble0 with four levels of reasoning budget, covering all scenarios from cloud to edge. The flagship Emperor512 has a reasoning budget of 512 units, 8-10 times that of current models, and may become the core of GPT-5.2.