# Towards a universal meta-optics solver via large language models

arXiv: https://arxiv.org/abs/2608.26417 · HTML: https://mlsanity.com/paper/2608.26417/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 9 of its 47 references that are in the map.
- First posted on arXiv 26 Aug 2026.

## Builds on (most cited first, 9 of 9)

- [Deep neural networks for the evaluation and design of photonic devices](https://mlsanity.com/paper/2007.00084/) (2020)
- [A multi-agentic framework for real-time, autonomous freeform metasurface design](https://mlsanity.com/paper/2503.20479/) (2025)
- [OptoGPT: A Foundation Model for Inverse Design in Optical Multilayer Thin Film Structures](https://mlsanity.com/paper/2304.10294/) (2023)
- [An Agentic Framework for Autonomous Metamaterial Modeling and Inverse Design](https://mlsanity.com/paper/2506.06935/) (2025)
- [Learning Electromagnetic Metamaterial Physics With ChatGPT](https://mlsanity.com/paper/2404.15458/) (2024)
- [Chat to Chip: Large Language Model Based Design of Arbitrarily Shaped Metasurfaces](https://mlsanity.com/paper/2509.24196/) (2025)
- [A Self-Evolving Agentic Framework for Metasurface Inverse Design](https://mlsanity.com/paper/2604.01480/) (2026)
- [Data driven approaches in nanophotonics: A review of AI-enabled metadevices](https://mlsanity.com/paper/2510.00283/) (2025)
- [A Comprehensive Review of Large Language Models for Nanophotonics: From Surrogate Modeling to Autonomous Design](https://mlsanity.com/paper/2608.18279/) (2026)

## Abstract (arXiv)

Metasurface design increasingly requires fast models that can operate across structurally distinct device families, rather than retraining a separate surrogate for every geometry class. Conventional neural network surrogates often depend on fixed-dimensional descriptors, family-specific output formats, and repeated architecture tuning, which limits their scalability across heterogeneous meta-atoms. Here, we present a unified large language model (LLM) workflow for multi-family metasurface modeling and inverse-design. Geometries, design parameters, and optical response channels were converted into a shared instruction-following text format and used to fine-tune Gemma-2-9B across 8 metasurface families. Compared with single-family baselines, the joint model simultaneously predicted the optical responses of all metasurface families while reducing the MSE for each family by an average of 56.5%. The same representation was also used for inverse design. These results show that a shared sequence-based LLM interface can provide a practical route to cross-family metasurface design while reducing the need for task-specific surrogate architectures.