# Sci-Rho: A Multilingual Visually-Grounded Symbolic Benchmark for STEM Problems

arXiv: https://arxiv.org/abs/2606.08034 · HTML: https://mlsanity.com/paper/2606.08034/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 18 of its 24 references that are in the map.
- First posted on arXiv 6 Jun 2026.

## Builds on (most cited first, 18 of 18)

- [Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answering](https://mlsanity.com/paper/2209.09513/) (2022)
- [MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts](https://mlsanity.com/paper/2310.02255/) (2023)
- [GSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models](https://mlsanity.com/paper/2410.05229/) (2024)
- [Large Language Models Are Not Robust Multiple Choice Selectors](https://mlsanity.com/paper/2309.03882/) (2023)
- [DynaMath: A Dynamic Visual Benchmark for Evaluating Mathematical Reasoning Robustness of Vision Language Models](https://mlsanity.com/paper/2411.00836/) (2024)
- [Benchmarking Vision Language Models for Cultural Understanding](https://mlsanity.com/paper/2407.10920/) (2024)
- [Words or Vision: Do Vision-Language Models Have Blind Faith in Text?](https://mlsanity.com/paper/2503.02199/) (2025)
- [PhysReason: A Comprehensive Benchmark towards Physics-Based Reasoning](https://mlsanity.com/paper/2502.12054/) (2025)
- [MM-MATH: Advancing Multimodal Math Evaluation with Process Evaluation and Fine-grained Classification](https://mlsanity.com/paper/2404.05091/) (2024)
- [VGRP-Bench: Visual Grid Reasoning Puzzle Benchmark for Large Vision-Language Models](https://mlsanity.com/paper/2503.23064/) (2025)
- [Fool Your (Vision and) Language Model With Embarrassingly Simple Permutations](https://mlsanity.com/paper/2310.01651/) (2023)
- [SciVerse: Unveiling the Knowledge Comprehension and Visual Reasoning of LMMs on Multi-modal Scientific Problems](https://mlsanity.com/paper/2503.10627/) (2025)
- [AgroBench: Vision-Language Model Benchmark in Agriculture](https://mlsanity.com/paper/2507.20519/) (2025)
- [True Multimodal In-Context Learning Needs Attention to the Visual Context](https://mlsanity.com/paper/2507.15807/) (2025)
- [Grounding Multilingual Multimodal LLMs With Cultural Knowledge](https://mlsanity.com/paper/2508.07414/) (2025)
- [Refusal Falls off a Cliff: How Safety Alignment Fails in Reasoning?](https://mlsanity.com/paper/2510.06036/) (2025)
- [EngTrace: A Symbolic Benchmark for Verifiable Process Supervision of Engineering Reasoning](https://mlsanity.com/paper/2511.01650/) (2025)
- [Vision Language Models are Confused Tourists](https://mlsanity.com/paper/2511.17004/) (2025)

## Abstract (arXiv)

Symbolic benchmarks have emerged as a key approach to assess model robustness under minor modifications to STEM-related questions. However, existing symbolic benchmarks mostly remain limited to mathematical reasoning, lack visual grounding, and are predominantly in English. In this work, we introduce Sci-Rho (Science Rhobustness), a dynamic benchmark for visually-grounded STEM problems spanning five subjects and seven languages, comprising 4,242 problem templates (606 per language) crafted by domain experts, including Olympiad medalists. Each template is implemented as executable Python code that generates diverse but equivalent problem instances by varying numerical values, visual patterns, geometric shapes, color schemes, and function types, resulting in 42,420 instances in total, each paired with reasoning steps and ground-truth solutions. We evaluated 17 state-of-the-art VLMs and discovered a noticeable gap between worst-case accuracy (defined as the proportion of problem templates that a model answers correctly across every generated variation) and average accuracy. We also discovered that smaller models show noticeable performance degradation across languages, whereas proprietary and larger models remain robust. Step-level evaluation reflects this same trend, revealing a significant gap between average F1 and worst-case F1 scores. Finally, our inspection of attention heads of a VLM reveals substantial cross-lingual variation in the relative attention allocated to image tokens compared to text tokens. Our work highlights the importance of evaluation beyond static benchmarks as a metric to measure the quality of VLMs.