The AI Hub · 2026-07-06 · AI Boutique Team · Pricing

The subscription showdown: Claude, ChatGPT, Cursor and Google.

Four ladders, the same three rungs: a $20 sampler, a ∼$100 working tier, and a $200 ceiling. Which rung — and whose ladder — depends on the work.

The market has quietly standardised. Every provider now sells the same shape: a free tier to hook you, a $20 tier that’s superb for chat and rationed for agents, and one or two power tiers for people whose output pays for the seat. Prices below were checked on 6 July 2026 — they change often, so treat the shape as durable and the digits as perishable.

The price ladders, per person per monthBar length = price · scale to $200
Claude
Pro $20
Max 5× $100
Max 20× $200
ChatGPT
Go $8
Plus $20
Pro $200
Cursor
Pro $20
Pro+ $60
Ultra $200
Google
AI Pro $19.99
Ultra 5× $99.99
Ultra 20× $200
Checked 2026-07-06 · Google cut its top tier from $249.99 and added the $99.99 rung in May · ChatGPT Free is now ad-supported in the US

What each ladder is actually selling

Claude
ChatGPT
Cursor
Google
$20 buys
Frontier models + Claude Code CLI, ~5-hour rolling usage windows
Broadest toolkit — chat, voice, images, Codex agent bundled
AI-native IDE with ~$20 of model credits; agent mode burns them fast
Gemini 3 Pro, 1M-token context, 2TB storage, Workspace integration
Working tier
Max 5× ($100) — the default for daily Claude Code use
Nothing between $20 and $200 — the gap in the lineup
Pro+ ($60, ~$70 credits) — where daily agent users land
Ultra 5× ($99.99) — new in May, undercuts rivals' $100 rung
Coding agent
Claude Code — terminal-first, deepest single-codebase reasoning
Codex — cloud sandbox + CLI, notably token-efficient
The editor itself — multi-model (Claude, GPT, Gemini in one IDE)
Antigravity — multi-agent, bundled with AI Pro and up
Fine print
Limits shared between Claude chat and Claude Code; "20×" is a multiplier, not unlimited
Free tier carries ads (US); the showcase models sit behind $200
Credits are dollar-denominated; premium models drain them ~5× faster
Credit system is harder to predict; storage bundle flatters the price

Checked 2026-07-06

If your subscription stack has grown by accretion rather than decision, that’s exactly the kind of line item our Reality Check prices up — including which of these four ladders your team should actually be standing on.

Key conclusions
01

Code all day → Claude Max 5× or Cursor Pro+. Terminal-first, one codebase, deep reasoning: Claude Code. Living in an editor, hopping between models: Cursor. Both working tiers pay for themselves in hours saved within the first week.

02

Mixed work, one seat → ChatGPT Plus. Still the broadest $20 in software. Just know the demo-reel models live at $200 — and there’s no middle rung when you outgrow Plus.

03

Live in Gmail and Docs → Google AI Pro. The 1M-token context and Workspace integration make it the best $20 for document-heavy teams — and the quiet May price cuts made its upper rungs the cheapest ceiling of the four.

04

Don’t stack all four. The standard set costs ≈$110 a month per person and we’ve never audited a business where four parallel subscriptions beat one well-chosen tier plus a free-tier backup. Pick one primary, upgrade on evidence.

05

Match the tier to the loaded hour, not the sticker. A $100 upgrade that saves a senior person three hours a month is free. Forty $20 seats nobody opens are not. Read the usage dashboards before the renewal, not after.

Book an AI Reality Check

Ready to make the AI you've already paid for pay you back?