SWE-INTERACT Benchmark

What is SWE-INTERACT?

SWE-Interact evaluates coding agents on multi-turn, interactive software engineering tasks where a simulated user provides vague instructions, reveals…

Released

2026-06-29

Evaluates

Agents & Tool Use, Code, Software & AI Compute

Openness

unknown

Importer

Claire Radar

Review status

unreviewed

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2606.30573

SWE-INTERACT paper and code

• Paper https://arxiv.org/abs/2606.30573
• Code https://github.com/scaleapi/SWE-Interact

Which sources cite SWE-INTERACT?

• SWE-INTERACT

Related Agents & Tool Use, Code, Software & AI Compute benchmarks

• $\tau^\tau$-Bench: An Environment for End-To-End, Realistic Agent Construction
• A3S-Bench
• Active-SWE
• AdaPlanBench
• Adaptive Adversaries
• ADeptS-Bench
• AeroCopilotBench
• Agent App Benchmark