Benchmarks suggest otherwise. Toqan's sql benchmark shows other models way up in the ranking compared to gpt-4o [1] Open Weight models specifically fine-tuned

Benchmarks suggest otherwise. Toqan's sql benchmark shows other models way up in the ranking compared to gpt-4o [1] Open Weight models specifically fine-tuned on sql generation and modification also rank pretty well compared to SOTA proprietary models. If you want to eval alternative models, check out sqleval [2] 1 https://prollm.toqan.ai/leaderboard/stack-unseen?type=conceptual,version,debugging,implementation,optimization&level=intermediate,advanced,beginner&tag=sql https://prollm.toqan.ai/leaderboard/stack-unseen?type=concep... 2 https://github.com/defog-ai/sql-eval https://github.com/defog-ai/sql-eval