SCDBench (Compilation Repair) - Compilation: leaderboard
SCDBench (Qin, Song and Gervais, 2026) tests LLM decompilation of 600 verified Ethereum smart contracts in three difficulty tiers: from EVM assembly and function-selector hints the model must write Solidity that compiles and behaves like the original. Percentage of the 600 contracts whose recovered Solidity compiles to the target artifact after the model gets one attempt to fix the compiler errors of its first answer; higher is better.
Metric: Compilable decompilations (%; the recovered Solidity compiles to the target artifact, of 600 contracts, after one same-model repair attempt from compiler errors). Source: arxiv.org. Saturation forecast: Around February 2027. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.3 Codex (High) | 90.33 | #114 (GPT-5.3 Codex) |
| 2 | Claude Opus 4.7 (High) | 69.17 | #75 (Claude Opus 4.7) |
| 3 | GLM-5 (Thinking) | 67.17 | #256 (GLM-5) |
| 4 | GLM-5 (Non-reasoning) | 64.33 | #256 (GLM-5) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=scdbench-compilation-repair-compilation · How It Works · Data refreshed daily, snapshot 2026-11-12.