# openai/SWELancer-Benchmark

This repo contains the dataset and code for the paper "SWE-Lancer: Can Frontier LLMs Earn $1 Million from Real-World Freelance Software Engineering?"

Repository: https://github.com/openai/SWELancer-Benchmark
Canonical: https://ross.abutalabs.com/products/swelancer-benchmark
License Family: other
Archived: true
Last push: 2025-07-18T02:02:20+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 32, release rhythm 35, longevity 40
- inputs: {"age_days": 561, "days_push": 412, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, archived, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1431, forks 135 (observed 2026-08-28T04:04:42.952395+00:00)

## What it is
SWE-Lancer is a benchmark dataset and evaluation harness measuring whether frontier LLMs can complete real-world freelance software engineering tasks worth up to $1 million in aggregate. The codebase has been merged into OpenAI's preparedness repository, so this repo now serves as a pointer to the active project.

## Use cases
- evaluate llms on real-world software engineering tasks
- benchmark coding agents on freelance-style issues
- compare frontier models on paid software tasks
- run swe-lancer evaluations
- research llm coding capability
- measure ai performance on real github issues

## When to choose
- you need a rigorous benchmark of LLM software engineering ability on real paid tasks
- you are researching how frontier models handle realistic freelance coding work
- you want reproducible evaluation harnesses from OpenAI's preparedness project

## When to avoid
- you want an actively developed repo - use openai/preparedness instead
- you need a general-purpose coding assistant rather than an evaluation benchmark
- you lack the compute or API budget to run frontier model evaluations

## Facets
- artifact type: dataset
- maturity: maintenance
- function: benchmarking, llm-inference, machine-learning
- domain: large-language-models, artificial-intelligence, developer-tools, testing
- platform: python, cross-platform
- tags: benchmark, evaluation, llm-evaluation, software-engineering, freelance-tasks, openai, swe-benchmark, research, docker

## Member repositories
- openai/SWELancer-Benchmark (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:42.952395+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:37:08.414647+00:00, confidence not recorded.
  - readme: https://github.com/openai/SWELancer-Benchmark (fetched 2026-08-28T04:04:42.952395+00:00, sha dd642c1c6b58)
- Data as of 2026-08-30T08:39:29.467469+00:00.