Skip to content

SWE-bench Multimodal

A multimodal variant of SWE-bench that adds visual context such as screenshots and design mockups to software engineering issue descriptions.

7 models

Higher is better.

  1. 1Claude Opus 5.5Anthropic61.4
  2. 2Claude Opus 5Anthropic59.4
  3. 3Claude Mythos 5Anthropic54.9
  4. 4Claude Fable 5.1Anthropic54.7
  5. 5Claude Sonnet 5.5Anthropic54.3
  6. 6Claude Opus 4.8Anthropic38.4
  7. 7Claude Sonnet 5Anthropic28.1
Higher is better. Bar colour: blue proprietary, orange open weights, green other.

About this benchmark

Metric
Code patch generation with visual context
Tasks
Multimodal software engineering tasks
Authors
SWE-bench team (2025)

Scores are as reported by model developers or evaluators and may use different settings (for example reasoning effort or tools). Compare with care.