AI usage limits in 2026: what each plan really allows

Key findings

• None of the major chat apps publishes a fixed message count. Limits depend on the model, the length of the conversation and what you attach.
• Claude and Gemini refill on a 5-hour window up to a weekly cap; Grok refills weekly. OpenAI calls Free and Go text chats “unlimited subject to abuse guardrails” and says “limits apply” to Plus.
• Long conversations, file uploads, image or video generation, Deep Research and the strongest models all draw down the allowance faster. A clear prompt that works the first time is the cheapest way to stretch it.

What each plan gets

Official wording from each provider’s plan page or help center. Monthly US prices.

How much text fits in one conversation

The context window is how much the AI can “see” at once: your messages, its answers and your files. When a chat outgrows it, the oldest parts drop out of view.

What uses up your limit fastest

Google states it directly: media generation, Deep Research and the Pro model consume more of the Gemini allowance than ordinary chat. Anthropic and OpenAI describe the same pattern in their help pages: limits depend on the model, the length of the conversation and attachments.

Length matters because every new message makes the model reread the whole conversation. A chat that has gone back and forth ten times costs far more per message than a fresh one. If you keep correcting an answer, you pay for each correction and for everything before it.

In practice: start a new chat when you change topic, attach only the pages you need, use the strongest model only for hard tasks, and give enough detail in the first prompt that you don’t need three more to fix the result.