Methodology: Official AI subscription limits tracker
Official AI subscription limits tracker
- Conditions
What we track. Consumer and team plans of AI assistants — Claude, Gemini, Microsoft 365 Copilot, Mistral, Grok, GitHub Copilot and Perplexity — and, separately, the published API usage tiers of the Claude API, the OpenAI API and the Gemini API. ChatGPT plans are not covered yet; third-party sites are never used as a substitute for a vendor's own page.
Prices by country. A plan's price is recorded as a buyer sees it on the vendor's own page, from measurement points in the US, UK, Germany and Canada — in the page's own currency, with no conversion. Perplexity prices and limits are recorded as shown to a buyer in the United States. A price seen on fewer than two distinct dates is marked as a first reading until a later check on another date confirms it.
What counts as a value. The vendor's own wording of each usage limit on its public pricing or help pages, with the date we read it: reset windows, weekly limits, "N times more usage" multipliers, messages, file uploads, voice, image generation, coding agents, context window, credits, document storage, and the monthly price. A number is recorded only when the vendor states one; otherwise the wording itself is the value. For API tiers: the spend cap (per month, or per 10 minutes for the Gemini API), requests and tokens per minute for the model the page names, and what an account needs to reach the tier.
What it is not. These are the providers' published limits; your own account can be limited differently. We do not count messages or test how limits are enforced, and third-party sites are never a source.
When a value is Unknown. When the official page cannot be read, or the wording is missing, has changed or is ambiguous. A value we have not recorded is shown as Unknown, never as zero.
How often. Every page is read once a day.
AI Subscription Limits Watch. The share of tracked subscription plans that publish at least one usage limit as a number rather than only in words ("more usage", "extensive usage"), with the number of plans (N) and the date. API tiers are not part of it.
Version. Version 1 of this method, in effect since 29 September 2026; plans added 30 September 2026; API usage tiers added 11 October 2026; Perplexity plans and by-country reading added 17 October 2026.
- Coverage
- On : 28 AI subscription plans from 7 providers, and 11 API usage tiers.
- Region
- the United States