◇ BenchmarkDatabases
PostgreSQL SELECT reference: 40.7% fewer tokens.
Dense reference SQL: the honest low end of the savings range. Read through the Lyrenth index, the page becomes one clean AIDocument of 17,500 tokens instead of 29,509 tokens of raw HTML: 1.7x smaller for every agent that reads it. Numbers straight from the API's economics block, measured 2026-07-05.
Raw HTML · 29,509 tokens
AIDocument · 17,500 tokens
| Cost to ingest raw HTML | $0.089 |
| Cost to ingest the AIDocument | $0.052 |
| Saved, per read, per agent | $0.036 |
Basis: $3.00 per 1M input tokens (Claude Sonnet / GPT-4 Turbo class), as reported in the API's economics block. Source page: https://www.postgresql.org/docs/current/sql-select.html
Reproduce it
One call, same numbers.
Every AIDocument response carries its own economics block, so this benchmark reproduces itself:
curl -X POST https://api.lyrenth.com/v1/aidocument \
-H "Authorization: Bearer $LYRENTH_KEY" \
-H "Content-Type: application/json" \
-d '{"url":"https://www.postgresql.org/docs/current/sql-select.html"}'