How to Save ChatGPT Usage

ChatGPT plans set usage limits that depend on the model, the tools you turn on, and the length of your conversations. Bigger models, deep reasoning, agents, and image or video generation all use more.

One long chat on the biggest model is the quickest way to hit a limit.

Free and Go

Lower limits. Fine for lighter use.

Plus

$20 a month, including Codex and higher limits.

Pro

$100, $200, or $500 a month for heavier workloads.

Business

Per-seat plans that include admin controls.

Limits change often. Confirm your plan’s current limits in ChatGPT’s settings and help center.

GPT-6 Luna

Focused, repeatable tasks. Uses the least capacity.

GPT-6.1 Sol

Complex work and agentic tasks. Uses more.

GPT-6 Astra

The toughest problems. Uses the most.

Start on the smaller model. Step up only when the answer falls short.

• Low effort for quick edits and simple questions
• High effort for planning, analysis, and debugging
• Deep research modes reserved for real research projects

• One chat for each topic
• Summarize, then carry on in a new chat
• No reusing old chats for new tasks

A long chat sends its full history along with every new message.

• Give context, goal, and format in a single message
• Group related questions together
• Request the length you need
• Edit your message rather than sending corrections

Keep reusable files and instructions in a project. Let memory hold your preferences so you never restate them in each chat.

Web search, connected apps, and agent features add work to every request. Turn them on when a task needs them, not by default.

• Write a detailed prompt on the first try
• Edit an image rather than regenerating it
• Make rough drafts before final high-quality versions

• Pick GPT-6 Luna for focused coding tasks
• Drop reasoning effort for routine edits
• Keep AGENTS.md brief
• Open new chats for new tasks
• Check usage and limits with /status

Agents and always-on Dots can consume a lot of capacity. Give them narrow tasks and stop any run that heads the wrong way.

• Switch down to a smaller model
• Clear simple tasks while you wait
• Use another tool for light work
• Upgrade only if limits block real work week after week

Answer briefly, without preamble. Use bullet points and stay under [length]. If anything is unclear, ask one question rather than guessing. Context: [context]. Task: [task]. Format: [format].