Introducing a novel method to execute robust AI securely and effectively on your existing devices.
Singapore, Nov. 06, 2025 — Gradient, an AI research and development laboratory dedicated to constructing open intelligence infrastructure, has publicly released Parallax, a novel AI operating system. This system empowers individuals and teams to create, host, and utilize potent AI applications directly on their personal devices and networks.
Supported by prominent global investors including Pantera, Multicoin, and HSG (formerly Sequoia China), Gradient’s mission is to democratize access to sophisticated AI infrastructure for everyone, everywhere. Parallax transforms standard hardware – such as laptops, desktops, and GPUs – into a cohesive, adaptive network that functions as a singular, high-performance computing unit. It empowers users to deploy their own copilots, assistants, or creative tools locally, with the capability to scale effortlessly when additional computational power is required.
Extensive AI models, which typically demand costly cloud infrastructure, can now operate cooperatively across various devices, automating workload distribution and routing to ensure peak speed and efficiency. Upon its release, Parallax offers compatibility with more than 40 open models across Windows, macOS, and Linux platforms, functioning on both NVIDIA GPUs and Apple Silicon. In extensive tests involving 14 networked machines, Parallax exhibited substantial efficiency improvements – achieving up to 3.6 times greater throughput and 2.6 times reduced latency when compared to leading local hosting frameworks.
“Parallax offers the simplest method for hosting your own agents and applications – securely, efficiently, and independently of cloud services,” stated Eric Yang, Founder and CEO of Gradient. “The introduction of Parallax represents a pivotal moment for the AI industry. For years, advancement has been synonymous with larger models and expanded cloud infrastructure. Sovereign AI constitutes the subsequent stage: transparent systems that any user can operate, validate, and enhance. We are transitioning from leased intelligence to personally controlled intelligence, where individuals and teams develop, host, and refine AI according to their specific requirements.”
Parallax is conceived for the forthcoming generation of AI applications that can remain confidential, auditable, and open-source. As individuals, developers, and businesses progressively depend on AI for tasks such as coding, communication, design, and research, the capacity to operate these systems securely and autonomously has become indispensable.
The system maintains data and memory locally, leverages open-source tools, and monitors each computational step, enabling reproducible and auditable outcomes. With its networking layer now operational, and upcoming verification and multi-agent functionalities, Gradient is establishing the foundational infrastructure for a sovereign AI ecosystem capable of global expansion.
The company has also unveiled the for – a distributed reinforcement learning framework that separates training from inference processes by utilizing Parallax to achieve efficient, consistent inference capabilities across networks of machines.
Developers and researchers can currently access Parallax via and review the comprehensive technical paper at .
About Gradient:
Gradient is an AI research and development laboratory committed to fostering open intelligence through a fully decentralized infrastructure – OIS (Open Intelligence Stack) – which includes distributed training, serving, agentic systems, and further advancements.
We are addressing the most challenging and ambitious problem at the confluence of Blockchain and AI: How can high-quality LLMs be trained and served across the public internet? Is constructing an OpenAI competitor feasible on a permissionless, autonomous network? Our present OIS comprises Parallax for distributed serving, Echo for distributed reinforcement learning, and Gradient Cloud for enterprise solutions.
Supported by leading investors and a team of world-class researchers, Gradient is dedicated to unveiling further cutting-edge research that will pave the way for a future where intelligence can be structured, expanded, and refined by anyone, globally.
CONTACT: Aroma Kumar aroma@lunapr.io
