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

1011

Output Tokens

129

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 DISTINCT u.user_id, u.name, u.email, u.created_at
FROM users u
JOIN completions c ON c.user_id = u.user_id
WHERE u.role = 'student'
  AND u.created_at >= date_trunc('year', now())
  AND u.created_at <  date_trunc('year', now()) + INTERVAL '1 year';