SWE-Lancer moved out; this repo just forwards the mail

What it does This repository originally housed the dataset and code for OpenAI’s SWE-Lancer paper, which asks whether frontier LLMs can earn $1 million from real-world freelance software engineering tasks. The README now states that the entire codebase has been merged into openai/preparedness, leaving this location as a redirect. If you are looking for the benchmark itself, it now lives elsewhere.

The interesting bit The premise reframes model evaluation as a gig-economy stress test: can a model invoice like a senior contractor rather than merely pass a unit test? The actual tasks, payouts, and results have moved out along with the code, so none of that detail is visible here anymore.

Key highlights

• Accompanied the paper “SWE-Lancer: Can Frontier LLMs Earn $1 Million from Real-World Freelance Software Engineering?”
• 1,440 stars suggest the community is curious about LLMs as billable labor.
• All code and data now live in the openai/preparedness repository.
• This repo functions strictly as a forwarding address.

Caveats

• The README is a two-sentence redirect; no methodology, dataset details, or results remain at this URL.
• The repository currently contains no code, dataset, or runnable instructions.

Verdict The active benchmark now lives in openai/preparedness. This repository is useful only as a historical citation stub; developers looking to experiment will find nothing to run.