Personalized news feed, driven by a persona and implemented by a LangGraph swarm.
Meta and Apple collaborated to optimize the 816-billion-parameter Muse Spark Open multimodal model for two M5 Ultra Mac Studios via the MLX framework, claiming it's the most capable US open-weight model for complex software engineering and multi-step reasoning. The partnership, publicly supported by CEOs Zuckerberg and Ternus, marks a rare alliance between the rivals.
This matters because it demonstrates how large open-weight models can be efficiently deployed on local hardware, enabling powerful AI capabilities for development and engineering tasks without relying on cloud infrastructure.
Brendan O'Leary discusses how real-world AI adoption depends on building trust rather than chasing model benchmarks. Drawing from 40 trillion tokens of usage data, he explains the four rungs of the AI adoption ladder and analyzes key failure points in context construction, multi-model routing, and feedback loops to help engineering leaders successfully scale agentic workflows. By Brendan O’Leary
Nvidia on Thursday announced a few new additions to its Nemotron family: Nemotron 3.5 Super VL, a 120B vision-language model for The post Nvidia squeezed a 74B agent model into 24GB so it can run on your laptop appeared first on The New Stack .
Hackers accessed roughly 1,900 GitHub accounts, then used them to slip credential-stealing workflows into more than 30,000 repositories. The malicious The post Hackers hijacked 1,900 GitHub accounts. Then they went after 30,000 repos. appeared first on The New Stack .
The article announces a proposal to integrate Software Carbon Intensity (SCI) as a first-class signal in OpenTelemetry, based on a keynote at Observability Summit Europe 2026. It likely discusses the design of a standard for measuring and reporting carbon emissions from software, enabling developers to incorporate environmental impact into their observability practices.
Ramp and Stripe have built internal "vibe-coding" platforms—Glass and Harbor respectively—that let non-engineers create complex, data-connected applications using natural language. Modeled after tools like Lovable.dev, these platforms are built by small teams (1-2 engineers in weeks) and prioritize safe internal data exposure and adherence to company design libraries. The trend highlights a low-cost strategy for platform engineering teams to democratize internal tooling while upskilling engineers in AI infra.
The incident is likely 1st confirmation of mercenary spyware delivered by targeted ads.
Software supply chain security company Chainguard has announced two beta products for running and testing code written by developers and AI agents. Chainguard Workspaces provides remote development environments based on hardened microVMs, while Chainguard Checks moves continuous integration into the development loop. The company announced both products at its Assemble London event. By Matt Saunders
The decoupling of compute from storage has positioned Apache Iceberg as the de facto open table...
Muse Spark Open is available today and brings leading agentic capabilities to your desktop for the first time on Apple M5 Ultra Mac Studios. Today, we’re releasing the weights for Muse Spark Open, the most capable US open-weight model to date. Muse Spark Open is a natively multimodal model with 816B total parameters (42B active) and a 1M-token context window. Built for coding and agentic work, it excels at long-horizon tasks, complex software engineering, and multi-step reasoning. We believe the future is for everyone