AI Pricing Is Becoming UX
GPT-5.6 turns capability into a price ladder, while Codex Spark turns speed into a separate button. AI pricing is becoming part of the interface.
Read the pieceKahWee Teng · Staff software engineer
I work on frontend systems and developer experience, then write about what the diagrams leave out: migrations, performance, AI-assisted engineering, and the tradeoffs that make software last.

Bearings
Working terrain
Three recurring questions guide this journal: how systems change, how tools change the work, and what survives outside a demo.
React, TypeScript, performance, and framework changes measured in real implementation work.
Follow the frontend route 02 · PracticeAgents matter, but so do review habits, cost controls, context, and the harness around the model.
Follow the AI practice route 03 · FieldworkCloudflare, self-hosting, and home infrastructure observed where constraints are harder to hide.
Follow the infrastructure routeTrailhead
A short route through the archive: measured comparisons, difficult migrations, and lessons that still hold.
GPT-5.6 turns capability into a price ladder, while Codex Spark turns speed into a separate button. AI pricing is becoming part of the interface.
Read the pieceUpgrading TanStack Table v8 to v9 cuts rendering time by 28% on 5,000-row tables. The real migration effort lives in TypeScript generic boundaries.
Read the pieceMy Remix-to-React Router v7 migration touched 40 files and removed 591 net lines. Imports were easy; TypeScript took the time.
Read the pieceAdobe Animate kept the Flash timeline alive for HTML5, but the web moved on. Maintenance mode confirms the bridge could not hold.
Read the pieceFrom the trail
GPT-5.6 turns capability into a price ladder, while Codex Spark turns speed into a separate button. AI pricing is becoming part of the interface.
Upgrading TanStack Table v8 to v9 cuts rendering time by 28% on 5,000-row tables. The real migration effort lives in TypeScript generic boundaries.
React Router v8 keeps the framework-mode API familiar and moves v7's future flags into defaults. That restraint matters after an expensive v6-to-v7 migration.
Closed-model guardrails remain tied to the provider, even as comparable open weights spread. Public safety needs controls that survive that reality.
Models make awkward code cheaper to replace. Review should block future risk and leave private ideas of neat to automated conventions.
Grok Build produces useful patches, but its weekly limit changes how I use it. The harness and quota now matter more than the model.