AI Database Optimizer and SQL Optimization Dataset

AI Database Optimizer analyses and optimizes SQL queries for MySQL and PostgreSQL. Recommendations of large language models (GigaChat, YandexGPT, OpenAI-compatible, local models via Ollama) are accepted only after safety, schema, result-equivalence and benchmark checks.

The release includes SQL Optimization Dataset v1.2.0 (directory dataset/, CC BY 4.0): 944 measured records from eleven experiments, including repeated runs for answer stability, including a comparison of GigaChat-2, GigaChat-2-Pro, GigaChat-2-Max, Qwen2.5-Coder-7B and a rule-based rewriter, with original and candidate SQL, execution plans, per-run timings, equivalence checksums, outcome labels and full LLM prompts and responses.

Программа анализа и оптимизации SQL-запросов с проверкой рекомендаций языковых моделей и открытый датасет результатов экспериментов.