Atlas

Benchmarks

← All benchmarks

SWE-Lancer Diamond - IC SWE (payout earned)

Code · 2025-02-18

SWE-Lancer turns real Upwork freelance software-engineering jobs into end-to-end-tested tasks priced at their original payouts. This row is the public Diamond split of individual-contributor (IC SWE) implementation tasks, scored as the total dollar payout of the tasks the model solves. Higher is better.

Top models (higher is better)

ModelScore
GPT-5112000
GPT-5 Mini75000
GPT-5 Nano49000
o1 (2024-12-17)43958
GPT-4.134000
o3-mini33833
GPT-4.1 Mini31000
GPT-4.1 Nano9000
Loading Atlas data…