SemanticAlign-Bench - Numerical Precision: leaderboard
The numerical precision claims of SemanticAlign-Bench (SA-Bench; Beihang University, Shanghai Jiao Tong University and Peking University, 2026): whether an agent-written reproduction repository implements the hyperparameters, thresholds and architectural dimensions the paper states, scored by GPT-5.5 on a five-level rubric from 0 to 1. Higher is better.
Metric: Semantic alignment score (0-1; numerical-precision claims (hyperparameters, thresholds, architectural dimensions); 30 ICLR, ICML and NeurIPS 2025 papers with 1,491 atomic implementation claims; the agent reads the paper and writes a reproduction repository, and GPT-5.5 scores each claim against the code on a five-level rubric (0, 0.25, 0.5, 0.75, 1) without executing it; mean over the 30 paper-level scores). Source: arxiv.org. Saturation forecast: Around February 2027. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.6 | 0.38 | #161 |
| 2 | DeepSeek V4 Pro | 0.36 | #188 |
| 3 | Gemini 2.5 Flash | 0.34 | #453 |
| 4 | GPT-4o | 0.08 | #741 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=semanticalign-bench-numerical-precision · How It Works · Data refreshed daily, snapshot 2026-11-13.