# PrismaCoder: Pushing the Boundaries of Open Multimodal Code Intelligence through Decomposed Rubric Synthesis

> Trains visual-code models using separate code and rendered-output rubrics for data filtering and reinforcement learning. Relevant evaluations include rebuilding webpages, charts and vector graphics from images.

- Authors: Qiushi Sun; Jingyang Gong; Yang Liu; Qiaosheng Chen; Zhangyue Yin; Zixian Huang; Hanxu Hu; Jing Xiong; Fangzhi Xu; Haochen Tan; Haoli Bai; Lifeng Shang; Xiaoguang Li; Ben Kao; Qipeng Guo; Lingpeng Kong; Jiahui Gao
- Venue: arXiv
- Status: Preprint
- Publication / posting date: October 14, 2026
- Date source: First arXiv submission; no reviewed venue publication is established.
- First arXiv submission: October 14, 2026
- arXiv: 2610.20538
- Reviewed arXiv version: v1
- DOI: 10.48550/arxiv.2610.20538
- Sources reviewed: 2026-10-20
- Added to the catalog: 2026-10-20
- Original paper: <https://arxiv.org/abs/2610.20538>
- PDF: <https://arxiv.org/pdf/2610.20538v1>
- Project: <https://qiushisun.github.io/PrismaCoder-Home/>
- BibTeX: <https://awesomerea.org/papers/citations/paper-prismacoder.bib>
- Catalog page: <https://awesomerea.org/papers/paper-prismacoder/>
- Record JSON: <https://awesomerea.org/papers/paper-prismacoder/record.json>
- Chinese version: [简体中文](https://awesomerea.org/zh/papers/paper-prismacoder/index.md)

## Scope and limitations

Combines visual reconstruction with text-driven generation. Judge scores do not establish behavioral equivalence. Training and evaluation share some judge families; the human check uses 100 samples. Instruction n-gram filtering cannot exclude all pretraining exposure.

## Why it fits this collection

Provides a recent approach to reconstruction from rendered evidence.

## Publication and review notes

Work in progress. DataCite confirms the DOI despite the arXiv pending label. The project code link is pending; the dataset API returned HTTP 401 anonymously. Reviewed model inventories include 9B and 35B-A3B-SFT with MIT card metadata.

## Primary sources

1. <https://arxiv.org/abs/2610.20538v1>
2. <https://arxiv.org/html/2610.20538v1>
3. <https://api.datacite.org/dois/10.48550/arxiv.2610.20538>
4. <https://qiushisun.github.io/PrismaCoder-Home/>
5. <https://huggingface.co/collections/QiushiSun/prismacoder>
6. <https://huggingface.co/QiushiSun/PrismaCoder-9B/tree/d5743a565f9f9b55da0798c4636059074f21602a>
7. <https://huggingface.co/QiushiSun/PrismaCoder-35B-A3B-SFT/tree/e2df1e846cbfe5a2aebc860917b02c742007c62e>

## About this entry

Reviewed against the paper and its primary sources. Cite the original authors using the linked paper or BibTeX.
Original editorial content: MIT. Linked projects, papers and assets retain their own rights and licenses.