NexusCortex Just Beat Opus 4.8 – and It's Open Source (github.com)

NexusCortex, an experimental cognitive architecture developed in Go, has surpassed the capabilities of Opus 4.8 and is now available as an open-source project. This research initiative combines concepts from Sparse Distributed Representations (SDR), associative memory, and online learning to create a small cognitive-system prototype. Notably, NexusCortex isn’t aimed at replacing existing large language models (LLMs) or claiming artificial general intelligence (AGI); instead, it focuses on building foundational AI system primitives, offering valuable insights into low-level AI mechanisms such as memory storage and retrieval, attention, and continuous learning. Significantly, the architecture features ten neural region modules inspired by human cognition, including areas responsible for memory, reasoning, and emotion. Its implementation includes SDR-based attention mechanisms and sparse ternary compute, achieving efficient performance with weights packed at just 0.25 bytes per parameter. Unlike typical LLMs, this prototype emphasizes a more nuanced approach to cognitive processing, such as episodic memory consolidation inspired by sleep cycles. With a user-friendly web dashboard for real-time interaction and monitoring, NexusCortex serves both as a research tool and an educational framework for developers interested in the underlying mechanics of AI systems, further enriching the community’s understanding of cognitive architectures.