DBPlanBench uses LLMs to optimize physical query plans
james_y_zou · x · 2026-07-21
The post announces DBPlanBench, a framework for test-time optimization of physical query plans with LLMs.
- It targets Apache DataFusion and asks whether LLMs can make query execution faster.
- The paper frames traditional query optimization as relying on cost models and heuristics that often miss semantic correlations in schemas and queries.
- DBPlanBench represents plans in a compact serialized form, lets an LLM propose JSON edits, and then uses an evolutionary search loop to refine candidate plans over iterations.
- The attached image shows the paper title, authors, and the optimization workflow diagram.
The talk is scheduled for a Boston DataFusion meetup during VLDB 2026.
More from Research
- Project APE finds verifier reliability drops when papers contain multiple errors — soumitrashukla9 · 2026-07-22
- Project APE says verifier costs fell about 90x in a year as Chinese open models lead — soumitrashukla9 · 2026-07-22
- OpenAI-linked paper says capability RL can make models more reward-seeking — MariusHobbhahn · 2026-07-22
- Project APE builds its verifier benchmark from 100 AI-written papers with injected errors — soumitrashukla9 · 2026-07-22
- Paper proposes a CRED taxonomy and benchmark to measure research-error detectors — soumitrashukla9 · 2026-07-22
- OpenAI says long-horizon models need safety and alignment checks across full action sequences — rhiever · 2026-07-22