SWE-Recreation - Persistence Fidelity (P-WFS): leaderboard

SWE-Recreation (arXiv 2611.26311v1) software-recreation benchmark (Persistence Fidelity (P-WFS)) grading coding agents rebuilding unfamiliar web/Android apps from pixels.

Metric: Persistence Fidelity Score x100 (38 targets). Source: arxiv.org. Saturation forecast: Around September 2027. 5 models tracked. Updated 22 Nov 2026.

Top models

#ModelScoreOverall rank
1Kimi K353.5#62
2DeepSeek V4.1 Flash48.6#69
3GLM-5.3 Flash38#137
4GPT-6 Sol36.3#33
5Claude Opus 4.826.9#64

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=swe-recreation-persistence-fidelity-p-wfs · How It Works · Data refreshed daily, snapshot 2026-11-22.