# SWE-PolyVision: Benchmarking Cross-Image Abductive Reasoning for Repository-Level Software Engineering

arXiv: https://arxiv.org/abs/2609.29754 · HTML: https://mlsanity.com/paper/2609.29754/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 16 of its 52 references that are in the map.
- First posted on arXiv 24 Sep 2026.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)

## Builds on (most cited first, 16 of 16)

- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Program Synthesis with Large Language Models](https://mlsanity.com/paper/2108.07732/) (2021)
- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [Terminal-Bench: Benchmarking Agents on Hard, Realistic Tasks in Command Line Interfaces](https://mlsanity.com/paper/2601.11868/) (2026)
- [SWE-smith: Scaling Data for Software Engineering Agents](https://mlsanity.com/paper/2504.21798/) (2025)
- [SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?](https://mlsanity.com/paper/2509.16941/) (2025)
- [Multi-SWE-bench: A Multilingual Benchmark for Issue Resolving](https://mlsanity.com/paper/2504.02605/) (2025)
- [SWE-Bench+: Enhanced Coding Benchmark for LLMs](https://mlsanity.com/paper/2410.06992/) (2024)
- [SWE-PolyBench: A multi-language benchmark for repository level evaluation of coding agents](https://mlsanity.com/paper/2504.08703/) (2025)
- [SWE-bench Goes Live!](https://mlsanity.com/paper/2505.23419/) (2025)
- [Benchmarking Multi-Image Understanding in Vision and Language Models: Perception, Knowledge, Reasoning, and Multi-Hop Reasoning](https://mlsanity.com/paper/2406.12742/) (2024)
- [Agent-RLVR: Training Software Engineering Agents via Guidance and Environment Rewards](https://mlsanity.com/paper/2506.11425/) (2025)
- [SWE-bench-java: A GitHub Issue Resolving Benchmark for Java](https://mlsanity.com/paper/2408.14354/) (2024)
- [FailureMem: A Failure-Aware Multimodal Framework for Autonomous Software Repair](https://mlsanity.com/paper/2603.17826/) (2026)
- [MM-IssueLoc: A Controlled Benchmark for Evaluating Visual Evidence in Multimodal Repository-Level Issue Localization](https://mlsanity.com/paper/2607.15205/) (2026)
- [GALA: Multimodal Graph Alignment for Bug Localization in Automated Program Repair](https://mlsanity.com/paper/2604.08089/) (2026)

## Abstract (arXiv)

Current multimodal software-engineering benchmarks expose images as additional context, but do not test whether an agent can integrate evidence distributed across images into a verified repository-level repair. We present SWE-PolyVision, an executable benchmark of 92 real tasks from 36 open-source organizations, with 48 public tasks and 44 private holdouts. The release contains 402 static images and 6 videos, with at least two visual inputs per task. Each task pairs a fixed pre-fix repository with an isolated verifier and is evaluated under the supported conditions among three access modes: Text-only, Native Vision, and Tool-mediated Vision. Across eleven coding models, visual access changes which tasks are solved, but effects depend on both model and task. Two trace-linked Native Vision cases illustrate how complementary visual and textual clues can lead to source-localized, verified repairs; controlled interventions show that this conversion is not yet stable across inputs. SWE-PolyVision thus separates the availability of multi-image evidence from its successful use in repository-level repair, without treating patch success alone as proof of explicit reasoning.