Coding Horizon

Your $200 AI Coding Plan Is Mostly Paying for Waste

Every figure this video puts on screen, chased to the primary source that publishes it. Prices are per user per month in US dollars, on the individual plans, as published on 2026-08-09. Subscription pricing moves; check the vendor’s own page before quoting these.

The price ladder

The video walks a ladder from a free trial to two hundred dollars a month. Each rung is a real published tier.

Rung What it is Source
$20 Cursor Pro Cursor, Models & Pricing
$60 Cursor Pro Plus Cursor, Models & Pricing
$100 Claude Max, lowest tier Claude pricing
$200 Cursor Ultra Cursor, Models & Pricing

Cursor Ultra

Cursor publishes Ultra at $200 per month with 20x more usage than Pro, and describes it as answering power users who wanted more predictability than usage-based pricing.

Claude Max

Codex, and metering in credits

The video says agentic tools increasingly talk in credits, tokens, model rates and usage pools. That is how Codex’s own documentation describes it.

The wider market

The video shows that the top tier is not one vendor’s experiment. GitHub Copilot publishes Free at $0, Pro at $10, Pro+ at $39 and Max at $100 per user per month.

What a coding agent actually does

The video lists reading files, planning, calling tools, generating diffs, running commands and sometimes spawning multiple attempts, and says all of it consumes compute. This is the documented behaviour of the agent modes in the tools above rather than a measurement: both Cursor and Codex describe agents that read a codebase, call tools and run commands, and both bill the resulting tokens.

The free toolchain it is compared against

The video weighs a $200 monthly editor against the tools it replaced. The comparison uses tools that are free and open source at the individual level: the editor itself, the linter, formatter and type checker, version control, and the package manager. No paid figure is claimed for any of them.

Not checked