Every AI coding plan, compared
A coding plan is a monthly subscription that pays for the model usage behind a coding agent. They differ in price, in how usage is measured, in the windows that ration it, and in which models you get. Here they are side by side, using each vendor's own pages.
Competitor details were checked against their official pages on November 4, 2026. Prices and limits change; follow the links to confirm.
The plans side by side
Prices and limits as each vendor's own page stated them on November 4, 2026. Follow the links to confirm: these change often.
| Plan | Price a month | What you get | Windows | Models |
|---|---|---|---|---|
| Token Plans | $20 / $100 / $200 | $60 / $300 / $600 of usage at list price | None; monthly credit | Claude, GPT, Gemini, GLM, Kimi, DeepSeek |
| Claude Pro / Max / Ultra | $20 / $100 / $200 | Pro usage, 5× Pro, 10× Pro (relative; no token figure published) | 5-hour and weekly (Ultra: weekly only, for verified accounts) | Claude, with Fable on Max and Ultra under its own weekly limit |
| ChatGPT Plus / Pro, for Codex | $20 / from $100 | Plus: estimated local messages per five-hour period; ChatGPT and Codex share usage | Plus: 5-hour; Pro: no 5-hour limit currently | OpenAI models |
| GitHub Copilot Pro / Pro+ / Max | $10 / $39 / $100 | 1,500 / 7,000 / 20,000 AI credits (1 credit = $0.01) | None; resets on the 1st | GitHub's model list, inside Copilot |
| Z.ai GLM Coding Plan | From $18 (Lite) | Lite / Pro / Max: 2,000 / 12,000 / 28,000 credits per 5 hours, 10,000 / 60,000 / 140,000 per week | 5-hour and weekly | GLM-5.3 and GLM-5.3-Flash |
| OpenCode Go | $10 / $40 (Go Plus) | Request allowances per model | 5-hour, weekly and monthly | Open-weight models, plus Claude Haiku 5.5 and GPT 6 Luna |
| Ollama Cloud | $20 / $100 | $60 / $300 of usage credits a month | Monthly credit | Open models |
Cursor is not on the list because its plans don't pay for Claude or GPT: Cursor bills third-party models separately at API list price.
How to read the table
- Dollars beat multiples. "5× Pro" only means something if you know what Pro is; Anthropic doesn't publish that. A dollar amount at list price — Copilot's credits, Ollama's credits, Token Plans — tells you in advance what the month buys.
- Windows ration the month. A 5-hour or weekly window protects the rest of the month from a heavy day, but it also stops you mid-task. Monthly credit lets you spend the month in a week if the work calls for it.
- One lab or many. The GLM Coding Plan is the best value if GLM is all you want, and Claude Max is the only plan with Fable included. A Token Plan's advantage is breadth: every frontier model in one pool, in any agent.
What a Token Plan buys
| Plan | You pay / month | Usage at list price | ≈ tokens on Claude Opus 5.5 | ≈ tokens on Claude Sonnet 5.5 | ≈ tokens on GLM-5.3 |
|---|---|---|---|---|---|
| Solo | $20 | $60 | 48M | 96M | 210M |
| Pro | $100 | $300 | 240M | 480M | 1B |
| Max | $200 | $600 | 480M | 960M | 2.1B |
Token estimates assume coding-agent traffic: 85% cached input, 12% fresh input, 3% output.
Where we lose
The GLM Coding Plan and OpenCode Go give more raw tokens per dollar if you only want their own or open-weight models, and Claude-only users who use every Max window may get more Claude from Anthropic's plan. We don't claim otherwise. A Token Plan is for people who want Claude, GPT, Gemini, GLM, Kimi and DeepSeek on one bill, with no windows, and a number they can read.
Prices on this page come from the model catalog and were current on November 12, 2026.