decomp.dev was born in 2024 as a way to centralize matching decompilation progress and to help expand a community of talented, like-minded folks.
Historically, matching decompilation has been a brutally difficult, time consuming task, on the scale of years for any moderately-sized game with a number of active contributors. As we developed better tooling and shared knowledge around the intricacies of decompilation, the compilers, and the games themselves, progress began to accelerate.
AI's use in decompilation has been experimented with for years, with mixed results. Starting with the release of Opus 4.6 and GPT-5.4 in early 2026, it was clear that the models were being trained on matching decompilation (they appeared to know our specific tools and how to use them without explicit guidance), and could achieve decent results inside of an established, clean codebase with exhaustive documentation. However, those results still required a significant amount of human cleanup and fixes for obvious issues or cheating (code that technically compiles to the same machine code, but could not be reasonably passed as what the original developers wrote), and the cost per task was prohibitively expensive relative to the results.
Recent models (like GPT-6-Astra) shifted things even further: they are significantly better at following complex directions, cross-referencing resources for symbols or debug information, and writing natural code with enough guidance. They are also exceptionally good permuters: they can write small scripts that test many different source code forms and combinations to find a match, even across files and headers to cover aspects like inlined helpers, signature changes, and operation ordering. In addition, their ability to remain coherent and focused across long-horizon tasks allows them to work continuously without going off the rails. Finally, they can achieve these better results faster and cheaper than earlier models could.
It is now clear that, with the right guidance, human input, and code reviews, AI can expedite the decompilation process in a significant way. What is also now clear is that the community aspect of decompilation has been broken. Our projects have always been open source to encourage contributions and to benefit the community as a whole. Bad actors have been abusing this in increasingly troublesome ways:
decomp.dev was originally designed to gamify decompilation: watching the progress go up was a fun way to stay motivated when every fraction of a percent took real human effort. Today, that same idea has warped decompilation into a hollow numbers game, exacerbated by folks who just want to burn tokens to make number go up or "port game to PC", all else be damned. While the increased attention has brought in genuine folks who want to learn and contribute meaningfully, I believe that decomp.dev serving as a public index and leaderboard is now causing more harm than good.
decomp.dev will continue to track existing projects and serve individual project pages and historical data, but the main index and adding new projects are disabled.
No. decomp.dev's repository has never had an open source license, and I continue to reserve all rights to the (entirely human written) code. Any replica must not use decomp.dev's code in any form.
I have long held the opinion that AI is not the issue by itself, and we now know it can help tremendously when used correctly. "Slop" comes in many forms, is hard to define, and even harder to moderate. As a sole hobbyist, I don't have the energy (or desire) to define what makes a "good" decompilation and enforce that consistently across the hundreds of projects on the site.
As this era of decompilation comes to a close, I want to thank everyone who's been involved and contributed to our projects, communities and tools. The people and passion behind these projects have always been what's made them fun and rewarding.