What should I use for this?

Neutral guidance mapping tasks to a model strength, a recommended platform, and a realistic budget. This is editorial judgement — versioned, dated, and labeled with confidence, not a claim that any model is definitively “best.”

Model strength, in plain terms

Light

Small edits, boilerplate, copy, simple scripts.

Standard

Most implementation, moderate debugging, APIs, tests.

Strong

Hard debugging, architecture, multi-file refactors.

Maximum

Deep architecture, large codebase analysis, long agents.

Pick a task

Build a SaaS MVP

strong

Ship a small but real SaaS product from scratch.

Build a Chrome extension

standard

A focused browser extension with a few features.

Debug a hard problem

strong

Track down a subtle, multi-file bug.

Refactor a large codebase

maximum

Restructure a big codebase safely.

Write boilerplate

light

Repetitive scaffolding, config, and copy changes.

The efficient AI build stack

1.Plan with Strong.
2.Build with Standard.
3.Use Light for repetitive edits.
4.Escalate difficult debugging to Strong.
5.Review with Strong.
6.Use Light or Standard for cleanup and documentation.