Europe's Ariane 6 rocket successfully launched for the first time Thursday, while OpenAI deployed a new coding model on non-Nvidia hardware, and Google Chrome introduced a new web standard for AI agents. These developments signal significant advancements in space exploration, artificial intelligence, and web technology.
The Ariane 6 rocket, carrying 32 spacecraft for Amazon's satellite broadband constellation, lifted off from the Guiana Space Center in South America at 11:45 am EST (16:45 UTC), according to Ars Technica. This marked the first launch of the Ariane 64 configuration, utilizing all four of the rocket's solid rocket boosters and generating over 3.4 million pounds of thrust.
Meanwhile, OpenAI released its GPT-5.3-Codex-Spark coding model, designed to run on chips from Cerebras, as reported by both Ars Technica and VentureBeat. This move represents OpenAI's first production AI model to operate outside of Nvidia hardware. The model reportedly delivers code at over 1,000 tokens per second, approximately 15 times faster than its predecessor. "Cerebras has been a great engineering partner, and we're excited about adding fast inference as a new platform capability," said Sachin Katti, head of OpenAI, according to Ars Technica. This partnership comes at a time when OpenAI is navigating a complex relationship with Nvidia, its primary chip supplier, as noted by VentureBeat.
In other AI news, Nvidia researchers developed a technique called dynamic memory sparsification (DMS) that can reduce the memory costs of large language model reasoning by up to eight times, according to VentureBeat. This method compresses the key value (KV) cache, the temporary memory LLMs use, without degrading the model's intelligence. Experiments show that DMS enables LLMs to "think" longer and explore more solutions without the same memory constraints.
Google Chrome also introduced WebMCP (Web Model Context Protocol) in an early preview, as reported by VentureBeat. This new web standard, developed jointly by Google and Microsoft, aims to transform every website into a structured tool for AI agents. WebMCP allows AI agents to understand and interact with websites more efficiently, potentially ending the current reliance on methods like scraping raw HTML and using screenshots.
Discussion
AI Experts & Community
Be the first to comment