Discover quality AI tools
Bring together the world’s leading artificial intelligence tools to help improve work efficiency
Articles published by ScottEvans
DeepSeek released the stable V4-Flash model (V4-Flash-0731) with enhanced agent capabilities and instruction following, now comparable to top closed-source models. The National Supercomputing Internet Platform launched its API and model downloads, offering one-click access and a Notebook environment for fine-tuning. Over 1,700 open-source models are available, enabling multi-model comparison and switching.
DeepSeek released the stable V4-Flash model (V4-Flash-0731) with enhanced agent capabilities and instruction following, now comparable to top closed-source models. The National Supercomputing Internet Platform launched its API and model downloads, offering one-click access and a Notebook environment for fine-tuning. Over 1,700 open-source models are available, enabling multi-model comparison and switching.
Anthropic's Claude AI exposed user session links due to a technical misconfiguration: conflicting robots.txt and X-Robots-Tag settings allowed Google and Bing to index shared URLs, revealing sensitive data like wallet keys and company tasks. Anthropic has since blocked crawling, but cached links persist.
Anthropic's Claude AI exposed user session links due to a technical misconfiguration: conflicting robots.txt and X-Robots-Tag settings allowed Google and Bing to index shared URLs, revealing sensitive data like wallet keys and company tasks. Anthropic has since blocked crawling, but cached links persist.
Zhizhu founder Tang Jie teased the next-generation GLM model as an epic plus upgrade, surpassing prior historic leaps. The new model, likely GLM-5.5 or GLM-6, may exceed 1 trillion parameters, potentially reaching 1.6 trillion, rivaling top US models. This follows GLM-5.2 achieving Opus-level AI coding with only 74B parameters. Competition intensifies as DeepSeek V4 release remains uncertain.
Zhizhu founder Tang Jie teased the next-generation GLM model as an epic plus upgrade, surpassing prior historic leaps. The new model, likely GLM-5.5 or GLM-6, may exceed 1 trillion parameters, potentially reaching 1.6 trillion, rivaling top US models. This follows GLM-5.2 achieving Opus-level AI coding with only 74B parameters. Competition intensifies as DeepSeek V4 release remains uncertain.
Ji Meng AI launched the Seedance 2.0 VIP version with native 4K video output, rendering directly at 4K resolution during generation for richer detail in hair, textures, and lighting. Unlike traditional upscaling, this meets professional needs in film, branding, and advertising, highlighting the trend of domestic AIGC tools converging on high-quality commercial applications.
Ji Meng AI launched the Seedance 2.0 VIP version with native 4K video output, rendering directly at 4K resolution during generation for richer detail in hair, textures, and lighting. Unlike traditional upscaling, this meets professional needs in film, branding, and advertising, highlighting the trend of domestic AIGC tools converging on high-quality commercial applications.
OpenAI accelerates its first AI agent smartphone launch to early 2027, a year ahead of schedule, to boost its IPO prospects. The device will feature a custom MediaTek Dimensity 9600 chip using TSMC's N2P process, dual NPU architecture, and hardware-level security. It aims for 30 million shipments by 2028.
OpenAI accelerates its first AI agent smartphone launch to early 2027, a year ahead of schedule, to boost its IPO prospects. The device will feature a custom MediaTek Dimensity 9600 chip using TSMC's N2P process, dual NPU architecture, and hardware-level security. It aims for 30 million shipments by 2028.
Tencent open-sources its HY-Embodied-0.5-X multimodal AI model for robotics. Optimized for real-world tasks, it enhances robot capabilities in spatial reasoning, fine operations, and executing complex instructions. The model is designed for practical applications like household services and desktop operations.
Tencent open-sources its HY-Embodied-0.5-X multimodal AI model for robotics. Optimized for real-world tasks, it enhances robot capabilities in spatial reasoning, fine operations, and executing complex instructions. The model is designed for practical applications like household services and desktop operations.
Investor Michael Burry escalates his campaign against Nvidia betting over one billion on an AI bust and alleging financial misconduct Nvidia refutes his claims as Burrys new Substack rapidly gains followers raising questions if his criticism could trigger the market collapse he predicts
Investor Michael Burry escalates his campaign against Nvidia betting over one billion on an AI bust and alleging financial misconduct Nvidia refutes his claims as Burrys new Substack rapidly gains followers raising questions if his criticism could trigger the market collapse he predicts
California Senate approved SB 53, mandating AI safety transparency from large firms, creating whistleblower protections, and launching CalCompute. Governor Newsom must now decide; firms under $500M face lighter rules. Opposed by tech lobbyists, backed by Anthropic.
California Senate approved SB 53, mandating AI safety transparency from large firms, creating whistleblower protections, and launching CalCompute. Governor Newsom must now decide; firms under $500M face lighter rules. Opposed by tech lobbyists, backed by Anthropic.
