SCDBench (Compilation Repair) - Perfect Contracts: 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 decompiled perfectly, every public function recovered and consistent with no spurious function, after the model gets one attempt to fix the compiler errors of its first answer; higher is better.
Metric: Perfectly decompiled contracts (%; every public function recovered and consistent and no spurious function, of 600, after one same-model repair attempt from compiler errors). Source: arxiv.org. Saturation forecast: Around 2029. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 (High) | 7 | #75 (Claude Opus 4.7) |
| 2 | GPT-5.3 Codex (High) | 6.83 | #114 (GPT-5.3 Codex) |
| 3 | GLM-5 (Thinking) | 3 | #256 (GLM-5) |
| 4 | GLM-5 (Non-reasoning) | 1 | #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-perfect-contracts · How It Works · Data refreshed daily, snapshot 2026-11-12.