SQL

Natural language to SQL query generation evaluates text-to-query fidelity and schema reasoning. This task is particularly relevant for analytics chat assistants and simplified database interfaces.

Duration

26s

Input Tokens

1126

Output Tokens

428

Cost

$0.01

Context

Input

Which students joined this year and already finished a lesson?

Expected output

SELECT DISTINCT u.name FROM users u JOIN completions c ON u.user_id = c.user_id WHERE u.role = 'student' AND u.created_at >= DATE_TRUNC('year', CURRENT_DATE);

Model output

SELECT DISTINCT u.* FROM users u JOIN completions c ON u.user_id = c.user_id WHERE u.role = 'student' AND EXTRACT(YEAR FROM u.created_at) = EXTRACT(YEAR FROM CURRENT_DATE);