HiL-Bench Original Paper - SQL - ASK-F1: leaderboard
Metric: ASK-F1 (%). Source: static.scale.com. 4 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 | 57.5 |
| 2 | Gemini 3.1 Pro (Preview) | 44.5 |
| 3 | GPT-5.4 | 32.1 |
| 4 | GPT-5.3 Codex | 22.3 |
Interactive version: theaggregate.ai/benchmark?slug=hil-bench-original-paper-sql-ask-f1 · How It Works · Data refreshed daily, snapshot 2026-09-19.