Neural Computers

2026-04-07

AI Summary

This work proposes a neural computer (NC)—a novel paradigm that treats the model itself as a running computer, unifying computation, memory, and I/O within a learnable runtime state. Addressing the limitations of traditional computing, which relies on explicit programs and struggles to integrate these components cohesively, and overcoming the lack of reusable, reprogrammable execution mechanisms in existing agents, the NC learns input–output trajectories to generate screen frames directly from instructions, pixels, and user actions in CLI/GUI environments—without access to underlying program states. Leveraging a video generation architecture, the approach learns interface primitives and, for the first time, conceptualizes the model as a universal computer. Preliminary experiments demonstrate the feasibility of fundamental interface primitives such as I/O alignment and short-term control, laying the groundwork for a complete neural computer (CNC).

Abstract

We propose a new frontier: Neural Computers (NCs) -- an emerging machine form that unifies computation, memory, and I/O in a learned runtime state. Unlike conventional computers, which execute explicit programs, agents, which act over external execution environments, and world models, which learn environment dynamics, NCs aim to make the model itself the running computer. Our long-term goal is the Completely Neural Computer (CNC): the mature, general-purpose realization of this emerging machine form, with stable execution, explicit reprogramming, and durable capability reuse. As an initial step, we study whether early NC primitives can be learned solely from collected I/O traces, without instrumented program state. Concretely, we instantiate NCs as video models that roll out screen frames from instructions, pixels, and user actions (when available) in CLI and GUI settings. These implementations show that learned runtimes can acquire early interface primitives, especially I/O alignment and short-horizon control, while routine reuse, controlled updates, and symbolic stability remain open. We outline a roadmap toward CNCs around these challenges. If overcome, CNCs could establish a new computing paradigm beyond today's agents, world models, and conventional computers.

Problem

• Neural Computers
• Completely Neural Computer
• learned runtime
• I/O alignment
• symbolic stability

Innovation

• Neural Computers
• learned runtime
• I/O alignment
• Completely Neural Computer
• program-free execution