← Back to News
Published on 6/1/2026

NVIDIA Launches NemoClaw and OpenShell for Secure Development

Primary Release
NVIDIA NemoClaw / OpenShell
Model Efficiency
TurboQuant (QVAC SDK 0.12.0)
New Open Source Model
Mellum2 (12B)

Daily Highlights

  • NVIDIA NemoClaw & OpenShell: NVIDIA debuted the NemoClaw framework today, a blueprint designed to help developers build orchestration layers for long-running digital coworkers. Complementing this is the OpenShell secure runtime, developed alongside Microsoft, Canonical, and Red Hat, which provides a containerized environment with native security primitives to manage privacy and policy controls for automated workflows.
  • Tether Releases TurboQuant: Tether AI Research Group announced the production release of its open-source TurboQuant implementation. Integrated into the QVAC SDK 0.12.0, this memory compression algorithm enables developers to run larger models and longer context windows on consumer-grade hardware, including laptops and edge devices, by optimizing memory usage.
  • SkipLabs Launches Skipper: SkipLabs introduced Skipper, a closed-loop coding tool designed to translate prompts into validated, running services. Built by the creator of Facebook's Hack language, Skipper acts as an architectural substrate that decomposes tasks, routes them to appropriate models, and handles the generation and validation pipeline to produce finished software.

Niche Project of the Day

Mellum2 (by JetBrains): Released today under the Apache 2.0 license, Mellum2 is a 12B parameter model engineered specifically for production software engineering systems. It is optimized for low-latency routing, summarization, and intermediate reasoning steps, offering developers a high-performance, cost-efficient alternative for integrating model-driven logic directly into their own infrastructure.

Key Takeaways

  • NVIDIA's NemoClaw and OpenShell provide a standardized, secure foundation for building complex, long-running automated workflows.
  • Tether's TurboQuant brings memory-efficient AI to local and edge devices, reducing dependency on cloud-based inference.
  • SkipLabs' Skipper focuses on the 'last mile' of development by turning prompts into validated, production-ready services.