Today, we announced plans to open source Blockhouse and launched the nFOX Platform, helping enterprises run open AI models on infrastructure they control.
Containment at the kernel level
Blockhouse restricts AI workload privileges by default. Enterprise policies determine approved access, while attempts to cross the sandbox boundary are blocked and recorded.
It supports vLLM, SGLang, and llama.cpp, with model integrity checks and an audited communication channel called Porthole.
Manage your private AI fleet
The nFOX Platform brings model evaluation, deployment, and monitoring into one enterprise-controlled console. Teams can approve models, serve them through OpenAI-compatible endpoints, and review integrity checks and containment events across their GPU infrastructure.
Availability
The nFOX Platform is available now. Blockhouse is in beta for qualified enterprise evaluations, with an open source release planned so developers and security teams can inspect its implementation.
Read the full announcement on EIN Presswire or contact us for a demonstration.