Roundtrip Benchmark

What is Roundtrip?

A roundtrip benchmark that scores natural-language code descriptions by regenerating code from them and measuring test pass fraction against original tests…

Released

2026-09-25

Evaluates

Code & Software, Software & AI Compute, cs.SE

Openness

unknown

Importer

Claire Radar

Review status

ai-reviewed

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2609.31587

Roundtrip paper and code

• Paper https://arxiv.org/abs/2609.31587
• Code https://github.com/haw-ai-i/roundtrip

Which sources cite Roundtrip?

• Roundtrip

Related Code & Software, Software & AI Compute, cs.SE benchmarks

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