Show Notes
Subquadratic is beginning to back up its ambitious claims with benchmarks and third-party validation for its SubQ 1.1 Small model, which uses its proprietary Sparse Attention (SSA) architecture to dramatically improve long-context performance. Rather than comparing every token to every other token, SSA selectively processes relationships, enabling near-linear scaling while maintaining high accuracy across context windows of up to 12 million tokens. The company reports near-perfect retrieval performance, competitive coding and reasoning benchmarks, and compute savings of up to 1,000x at maximum context lengths.
More from The New Stack Podcast
Kubernetes has become the dominant container orchestration platform, used by more than 80% of container users, but widespread adoption hasn’t eliminated tensions between developers and operations teams. Developers value Kubernetes for its self-service capabilities and ability to move quickly, while operations teams prioritize stability, governance and controlled change. In this episode, HPE Fellow David Estes explains how those competing priorities are becoming more challenging as applications spread across hybrid and edge environments and AI inference workloads increase demand for GPU scheduling and rapid capacity scaling.
Open-source technology has played a foundational role in the AI boom, giving companies the flexibility to scale unprecedented computing workloads across providers and hardware. OpenAI, for example, used Kubernetes as early as 2018 to balance compute across its own data centers, AWS and Azure — years before ChatGPT brought generative AI into the mainstream.
Dynatrace sells software that monitors companies’ apps and infrastructure and wants its AI agents to make engineers’ jobs easier. An agent that gets in the way is a bad agent and should be dropped, said Sean O’Dell, a principal product marketing manager at the company.
For many developers, turning an AI-generated prototype into maintainable software requires more than generating code—it requires infrastructure, collaboration, testing and review. In this episode of The New Stack podcast, Bit Cloud founder and CEO Ran Mizrahi discusses Bit Cloud 2.0 and Hope, the company’s AI builder, and how they connect application creation with the work that follows.
CloudBees CEO Mo Plassnig is leading the CI/CD company through a major transformation as generative AI reshapes software development. Returning to CloudBees eight years after joining through its acquisition of CodeShip, which he co-founded, Plassnig says the emergence of generative AI renewed his interest in DevOps and the opportunities ahead.