SWE-Marathon Benchmark

What is SWE-Marathon?

SWE-Marathon evaluates AI agents on 20 ultra-long-horizon software engineering tasks, each with a unique executable environment, a human-written reference…

Released

2026-06-05

Evaluates

Code, Code & Software, Software & AI Compute

Openness

unknown

Importer

Claire Radar

Review status

unreviewed

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2606.07682

SWE-Marathon paper

• Paper https://arxiv.org/abs/2606.07682
• Website https://swe-marathon.org/

Which sources cite SWE-Marathon?

• SWE-Marathon

Related Code, Code & Software, Software & AI Compute benchmarks

• Active-SWE
• AFT-Bench
• Agent App Benchmark
• Agent Retrieval Bench
• Agent Systems Benchmark (ASB)
• Agentic Evaluation in 2D Mazes
• Agentic Security Benchmark
• AgentInfraBench