● Live Database Benchmark · MongoDB 8.3.4 vs Postgres 16.14

How many tokens does an AI agent spend
building on MongoDB vs Postgres?
We measured. No hypotheses. No claims. Just data.

The world's most realistic benchmark for database AI-friendliness. A 40-entity enterprise schema. Real CLI agents. Live databases. Zero No claims, no hypotheses, no verdicts. Just raw data from real agents on real databases.

40
Entities in schema
30/30
Cells passed clean
2
Independent AI agents
100%
Live DB verification
The Result

MongoDB wins. Both agents agree.

We asked two independent AI coding agents (Codex gpt-5.6-terra and Claude Sonnet) to add a field to a 40-entity enterprise schema and update a query. Each agent was compared only to itself — never cross-CLI. Both spent significantly more tokens on Postgres than MongoDB.

Median tokens read across 5 clean repeats per cell. Lower is better.

No hypotheses. No verdicts. Just the raw numbers — you decide what they mean.

Zero Bias · 100% Legit

How we proved it's not rigged

Every benchmark claims to be fair. This one proves it with 10 independent defenses — any one of which would catch bias if it existed.

The Pipeline

How a benchmark cell runs — step by step

No mocks. No simulated agents. No estimated tokens. Every number comes from a real AI agent writing real code against a real database. vanilla mode, anti-cheat, within-agent comparison — zero bias.

The Money

What this costs at enterprise scale

If your team runs 1,000 AI database-tasks per month, here's the monthly token bill per database design — based on the measured medians. MongoDB saves significant money on every agent.

Assumptions: 1,000 tasks/month. Token cost estimated at $3/M input tokens (Claude) and $2/M input tokens (Codex). within-agent comparison only — absolute costs are not cross-comparable between agents because CLIs count tokens differently. vanilla mode: no skills, no code reviews, no subagents. anti-cheat: source-pattern scan, protected files, cheat-signal detection.

The Schema

A real enterprise schema — 40 entities, 6 domains

Based on research into real enterprise database complexity (SAP, Salesforce, Workday). 40 entities across 6 business domains. MongoDB stores them as 21 collections with embedded arrays. Postgres stores them as 40 normalized tables with foreign keys and CHECK constraints.

The Task

A real enterprise day-to-day change request

The agent receives a change ticket — exactly what a developer gets in an enterprise. It must understand the existing schema, find where the field goes, write the migration, update the query, and not break anything.

Change Request: Add preferredPaymentMethod to accounts

"Add a preferredPaymentMethod field (string, optional, enum: credit_card / wire_transfer / ach / paypal) to the accounts entity. Update the getOrderSummary query to include the account's preferredPaymentMethod in the result."

  1. Read src/schema.mjs to understand the 40-entity schema
  2. Read src/accounts.mjs for the account model patterns
  3. Read src/queries.mjs for the order summary query
  4. Update the schema to add the new field to the accounts entity
  5. Update the account model to handle preferredPaymentMethod in create/update
  6. Update getOrderSummary to include preferredPaymentMethod in the result
  7. Run npm test until it passes — tests check regression AND new behavior
  8. Tests query the LIVE database directly — no mocks, no fakes
✓ Regression tested ✓ Live DB verified ✓ Schema evolution + query update Protected files: db.mjs, tests/