swe-bench/swe-smith

Coding

SWE-smith: NeurIPS 2025 toolkit for synthesizing unlimited SWE-bench-style task instances from any Python repository, plus released task instances and agent trajectories.

← Back to Registry

Run this task

CLI:

inspect eval inspect_harbor/swe_bench_swe_smith --model openai/gpt-5

Python:

from inspect_ai import eval
from inspect_harbor import swe_bench_swe_smith

eval(swe_bench_swe_smith(), model="openai/gpt-5")

Dataset information

Harbor registry swe-bench/swe-smith
Inspect task swe_bench_swe_smith
Latest digest sha256:e1ffe3cc3517754681749eda196485f51b38b4552340a48b12f4b6953b06aa86
Samples 100
Paper arxiv
Source https://github.com/SWE-bench/SWE-smith

See Task Parameters for the parameter set shared across all Harbor tasks.