II.
Benchmark overview
Reference · livebenchmark:swe-lancer
SWE-Lancer overview
OpenAI benchmark of >1,400 paid freelance software-engineering tasks scraped from Upwork, with associated bounties. Measures end-to-end agent ability to deliver shippable freelance work.
Attributes
displayName
SWE-Lancer
homepageUrl
kind
full-stack
targetsKind
AgentVersion
description
OpenAI benchmark of >1,400 paid freelance software-engineering tasks
scraped from Upwork, with associated bounties. Measures end-to-end
agent ability to deliver shippable freelance work.
Outgoing edges
applies_to1
- domain:software-engineering·DomainSoftware Engineering
covers2
- skill-area:autonomous-coding-engagement·SkillAreaAutonomous Coding Engagement
- skill-area:bug-fixing-from-issues·SkillAreaBug Fixing from Issue Descriptions
Incoming edges
None.