Benchmark RadarAI BENCHMARK PROFILE
ClarifyCodeBench
ClarifyCodeBench evaluates LLMs on clarifying ambiguous requirements for code generation through interactive dialogues. It includes manual annotations of ambiguity types, clarification questions, and ground-truth answers, with metrics for interaction quality.
- Released
- 2026-07-01
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
Code generation in practice involves underspecified requirements, but existing benchmarks assume perfect prompts. ClarifyCodeBench addresses this by measuring a critical yet underexplored capability, revealing that strong code generation does not imply effective requirement clarification.
Motivation
Large Language Models have emerged as programming assistants.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.