stack8s reports that hardware hacks for GPT-6 Astra have been working now after months of development.

We are tackling the problem of cloud platforms being expensive and difficult to manage, especially for startups leveraging AI/ML services, which can see costs rise dramatically. Many organizations don't have visibility into where their cloud spend is going, and inefficiencies lead to billions of dollars in wasted resources, with environmental impacts from idle compute. Vendor lock-in further complicates optimization, making it costly to migrate data and manage cloud expenses effectively. Our focus is on addressing these financial, environmental, and flexibility challenges.