A foundation model of numerical intelligence with cross-disciplinary generalization

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Chenghan Wu, Zongmin Yu, Liu Yang
• Venue: arXiv: cs.AI
• Published: 2026-07-30
• Collected: 2026-07-31
• URL: http://arxiv.org/abs/2607.28432v1
• PDF: https://arxiv.org/pdf/2607.28432v1
• DOI: UNKNOWN

Plain-Language Brief

The paper introduces UNified In-Context Operator Networks (UNICON), a model that demonstrates numerical intelligence across various disciplines. It is capable of inferring predictive relations from graph-based examples and applying them to queries within the same system, even in scientific and social systems unseen during training. The authors claim that UNICON can surpass state-of-the-art specialists in a discipline unseen during training when combined with language-model agents.

What the Document Says

• The paper presents UNified In-Context Operator Networks (UNICON), a foundation model of numerical intelligence.
• UNICON infers the predictive relation shared across graph-based examples from a system and applies it to queries from the same system.
• Across scientific and social systems, including those absent during training, UNICON performs similarly to specialist models without retraining.
• When combined with language-model agents, UNICON surpasses state-of-the-art specialists in a discipline unseen during training.
• The authors find that training-corpus diversity improves generalization to unseen disciplines.

Key Concepts Mentioned

• UNified In-Context Operator Networks (UNICON)
• numerical intelligence
• graph-based examples
• predictions
• scientific and social systems
• specialist models
• language-model agents

Claims Made by the Document

• UNICON is a foundation model of numerical intelligence.
• UNICON can generalize across disciplines, including those not seen during training.
• When combined with language-model agents, UNICON outperforms state-of-the-art specialists in unseen disciplines.

Evidence Used

• The authors present UNICON and demonstrate its performance across various disciplines.
• Results show similar performance to specialist models in scientific and social systems, including those absent during training.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Intelligence is commonly understood as the ability to acquire and apply knowledge, adapt to unfamiliar situations and solve new problems. Large language models exhibit this capacity by inferring task-relevant knowledge from textual context and applying it to new tasks. Yet intelligence need not be confined to language. For scientific and social systems, we need models that acquire and apply knowledge from numerical context-an ability we call numerical intelligence. Here we introduce UNified In-Context Operator Networks (UNICON), a foundation model that exhibits numerical intelligence across disciplines. Using graph-based examples from a system as context, UNICON infers the predictive relation shared across them and applies it to queries from the same system. Across scientific and social systems, including those from disciplines absent from training, the same model approaches specialist performance without retraining. Combining UNICON with language-model agents yields further gains, enabling it to surpass state-of-the-art specialists in a discipline unseen in training. We further show that training-corpus diversity improves generalization to unseen disciplines. Together, these results establish UNICON as a foundation model of numerical intelligence and position it as a building block for a broader ecosystem of artificial intelligence.