ClarifyCodeBench - Optimal Round Adherence (ORA): leaderboard
Metric: Optimal round adherence (0 to 1; Gaussian credit exp(-(n-K)^2/(2 sigma^2)) for using the annotated number of clarification rounds plus the final coding round, sigma = 0.425 K, on 419 ambiguous LiveCodeBench v6 tasks; at most 6 rounds, temperature 0). Source: arxiv.org. Saturation forecast: Around 2031. 7 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V3.2 (Non-reasoning) | 0.5 |
| 2 | GPT-4o | 0.33 |
| 3 | GPT-5 | 0.26 |
| 4 | Gemini 2.5 Flash (Non-reasoning) | 0.23 |
| 5 | Claude Sonnet 4.5 | 0.22 |
| 6 | Claude Sonnet 4.5 (Thinking) | 0.2 |
| 7 | Qwen 3 235B A22B (Non-reasoning) | 0.13 |
Interactive version: theaggregate.ai/benchmark?slug=clarifycodebench-optimal-round-adherence-ora · How It Works · Data refreshed daily, snapshot 2026-09-29.