Skip to content
TopicTracker
From HackerNewsView original
TranslationTranslation

DeepSWE Benchmark updated with GLM 5.2 and updated results for other models

The DeepSWE Benchmark has been updated to include GLM 5.2 and revised results for other models, providing a more current comparison of software engineering agent performance.

Background

- DeepSWE is a benchmark (standardized test) that measures how well AI coding agents can autonomously solve real-world software engineering tasks, like fixing bugs or implementing features.

Related stories

  • The author tests the new open-weights GLM 5.2 model against Gemini 3 Flash by having them play text adventure games within a fixed budget of $0.15 per attempt. A harness tracks achievements earned in each game to compare the models' performance.