PLSQLBench - Spider2 Multi-Turn (Test): leaderboard
Mean test pass at one across the turns of 63 held-out multi-turn Spider 2 conversations in PLSQLBench (2026), execution-based, in percent; higher is better.
Metric: Mean test pass@1 (%; the turns of the 63 held-out, database-disjoint multi-turn Spider 2 conversations, each turn conditioned on the model own earlier answers; direct generation of Oracle PL/SQL program units at temperature 0, executed on Oracle Autonomous Database 23ai; per-task fraction of unit tests passed by one solution, averaged over tasks). Source: arxiv.org. Saturation forecast: Around April 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 | 73.19 | #121 |
| 2 | Gemma 4 31B | 70.57 | #393 |
| 3 | GPT-5.6 Sol | 66.81 | #29 |
| 4 | Claude Opus 4.8 | 66.4 | #56 |
| 5 | GPT-5.4 Mini | 65.94 | #389 |
| 6 | Grok 4.3 | 65.46 | #332 |
| 7 | Gemini 2.5 Flash Lite | 56.78 | #958 |
| 8 | Llama 4 Maverick | 52.85 | #1105 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=plsqlbench-spider2-multi-turn-test · How It Works · Data refreshed daily, snapshot 2026-11-12.