Fort Lauderdale, Florida , March 02, 2026 (GLOBE NEWSWIRE) — Qubic has just released Neuraxon 2.0, a bio-inspired neural computation framework built on trinary logic and continuous-time processing, proposing an alternative architectural foundation for adaptive artificial intelligence systems.
Detailed in a newly published research paper, Neuraxon 2.0 moves beyond conventional binary perceptron models by introducing three-state neural units that operate across excitatory, neutral, and inhibitory dynamics. The framework is designed to process continuous input streams, enabling neural states to evolve over time rather than reset between discrete training cycles.
The model incorporates multi-timescale synaptic dynamics, neuromodulation-inspired regulation, and structural plasticity mechanisms aimed at supporting sustained real-time adaptation. According to the scientific research team, this approach is intended to more closely reflect biological neural organization while improving computational efficiency and resilience in non-stationary environments.
Alongside the paper, Qubic has released open-source code via its GitHub repository, as well as an interactive demonstration available on Hugging Face, allowing developers and researchers to examine and experiment with the architecture directly.
“Neuraxon 2.0 represents our effort to rethink neural computation from first principles,” according to David Vivancos & José Sánchez members of the Qubic scientific team “By introducing trinary state processing and continuous dynamics, we are exploring how adaptive systems can evolve in real time, rather than operate within fixed training boundaries.”
The research outlines how complemented trinary states may reduce iterative information loss in recurrent updates and how embedded synaptic modulation can enable learning across multiple temporal scales. The framework also examines hybridization with Qubic’s broader infrastructure research, where decentralized systems are evaluated as environments for hosting adaptive computational models.
The release of Neuraxon 2.0 forms part of Qubic’s open-science initiative, emphasizing transparency through peer-accessible documentation, executable code and reproducible experimentation. The project is positioned as foundational research rather than a commercial product launch, including a first real physical application NeuraxonMini with the integration of Neuraxon 2.0 as the “brain” of the Sphero Mini robot, a step forward in bringing “LiveAIs” instead of “DeadLLMs” to the real world.
About Qubic
Qubic is a decentralized infrastructure research initiative focused on advancing distributed computation and adaptive intelligence systems. The project explores novel consensus mechanisms, scalable network performance and bio-inspired computational models designed to operate in real-time environments. Through open research, public documentation, and community engagement, Qubic aims to contribute to the evolution of resilient and energy-efficient digital infrastructure.
