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

Input Tokens

979

Output Tokens

148

Cost

$0.00

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 u.name FROM users u JOIN enrollments e ON u.user_id = e.user_id JOIN completions c ON u.user_id = c.user_id WHERE u.role = 'student' AND u.created_at >= '2021-01-01' AND e.status = 'active' GROUP BY u.name;