Apple's Siri AI Finally Works in iOS 27 Beta
Apple's Siri AI, now functional in the iOS 27 beta, understands personal context and world knowledge, but its release feels anticlimactic after years of delays.
ARTICLES BY ALEX LINDGREN
LATEST NEWS
-
-
China's MiniMax H3 Becomes First Open Model to Top AI Video Ranking
China's MiniMax H3 video model ranks first in Video Editing and second in Text-to-Video on Artificial Analysis, marking the first time an open model leads an AI video benchmark.
-
Andrej Karpathy Tests Claude Opus 5 with Lord of the Rings Scene
Andrej Karpathy used Claude Opus 5 to generate 5,500 lines of code from a single 'Lord of the Rings' paragraph in about two hours, costing roughly $10.
-
Anthropic's Claude Opus 5 Generates Full 3D Game Worlds from Text Prompts
Anthropic's Claude Opus 5 creates fully functional 3D game worlds, including first-person shooters and Minecraft clones, using only text prompts, as shown in community tests.
-
ByteDance's Seedance 2.5 Creates 30-Second Video Clips with Audio
ByteDance's Seedance 2.5 generates 30-second video clips with built-in audio, surpassing Google's Gemini Omni Flash, which maxes at 10 seconds.
-
OpenAI Unveils Astra Model Family Solving Ten Math Problems
OpenAI announced its new Astra model family, which solved ten previously unsolved math problems, including non-sofic groups in group theory, in a report released on August 1, 2026.
-
xAI Unveils Grok-Imagine-Video-1.5 with References
xAI has launched Grok-Imagine-Video-1.5, its best video model yet, now supporting text, image, and voice references for generating up to 1080p video.
-
Google Earth retracts AI tool for fake satellite images
Google removed an AI feature from Google Earth after users created misleading images, including a fake Trump statue near the White House. The tool allowed AI-generated satellite modifications.
-
AWS Deploys Kimi K3 Model on SageMaker HyperPod and EKS
Moonshot AI's Kimi K3, a 2.8 trillion parameter model, is now deployable on AWS with two methods, as announced on July 27, 2026.
-
Deepseek Flash V4 Matches GPT-5.6 Luna at 60 Percent Lower Cost
Deepseek's Flash V4 model scores 50 on the Artificial Analysis Intelligence Index, nearly matching OpenAI's GPT-5.6 Luna at 60 percent lower cost.
-
Google DeepMind Unveils Gemini Robotics 2 for Versatile Robot Control
Google DeepMind has launched Gemini Robotics 2, a vision-language-action model capable of controlling robots from tabletop arms to humanoids, according to a July 31, 2026, report.
-
Chinese AI Researchers Gain Voice on X
Chinese AI researchers are increasingly using X to share their work, with over 30 Moonshot AI accounts active on the platform as of mid-2025.