OpenAI ↗ · 2025-02-18 18:00 · UTC+8
SWE-Lancer benchmarks economically meaningful engineering work
Company news OpenAI Source language: English
OpenAI introduced SWE-Lancer, a benchmark built around software engineering tasks with real economic value. Evaluation shifted from answer accuracy toward whether a model can complete deliverable work.
Excerpt
OpenAI introduced SWE-Lancer, a benchmark built around software engineering tasks with real economic value. Evaluation shifted from answer accuracy toward whether a model can complete deliverable work.
Short excerpt from the source — open the original for the full text.