IndustryCode - Python Sub-Problems: leaderboard

IndustryCode (2026) is a code-generation benchmark of hierarchical industrial programming problems in Python, C++, MATLAB and Stata, each main problem decomposed into sub-problems checked by test cases. This board is Pass@1 in percent on the Python sub-problems, each generated with the earlier steps as context; higher is better.

Metric: Pass@1 (%) on the Python sub-problems of the IndustryCode test split (106 main problems decomposed into 499 sub-problems; industrial Python, C++, MATLAB and Stata projects with test cases), each sub-problem generated with the main-problem description and the code of earlier steps as context; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 17 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Claude Opus 4.561.8#160
2Gemini 3 Pro59.7#145
3Claude Sonnet 4.558.6#261
4Qwen 3 Max56.6#415
5Gemini 2.5 Pro54.5#276
6Claude Haiku 4.553.4#583
7Seed 1.8Thinking53.4#252
8Kimi K2 (Thinking)51.8
9GLM-4.7Thinking50.3#380
10Qwen 3 Coder Plus50#720
11DeepSeek V3.2 Exp49.5#439
12Kimi K248.7#443
13GPT-5.248.5#185
14GPT-5.148.5#240
15Grok 4.1Non-reasoning47.6#462

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=industrycode-python-sub-problems · How It Works · Data refreshed daily, snapshot 2026-11-21.