ChatGPT: Adjusting Effort to Manage Usage Limits

Understanding the Intelligence Levels of ChatGPT

ChatGPT offers the ability to adjust the level of intelligence, thereby influencing the amount of thought put into a response before it is provided. A higher intelligence level allows the model to reason more extensively, improving the quality of responses for complex tasks. However, this increases response time and consumes usage limits more quickly. Conversely, a lower level speeds up responses and conserves resources.

The mobile application offers three intelligence levels, accessible based on the chosen subscription:

• Instant: For quick responses to simple queries.
• Thinking: For more elaborate reasoning when the instant mode is insufficient.
• Extended: For in-depth reflection, suitable for complex problems.

A Long Press to Adjust the Effort of Each Message

Adjusting the effort level does not require navigating through a menu. Simply long-press the send button after typing the message in the ChatGPT mobile app. A selector appears, allowing you to choose between the Instant, Thinking, or Extended levels before sending the message.

This method enables the customization of each message without altering the default model for the rest of the conversation. OpenAI favors this approach to automatically adapt the model to the complexity of the request.

Which Level to Choose to Preserve Usage Limits?

To optimize the use of ChatGPT, it is advisable to align the effort level with the complexity of the task. The Instant mode is sufficient for rephrasing, corrections, or factual questions. The Thinking and Extended modes are more suitable for in-depth analyses, complex code, or multi-step reasoning.

Unlike Claude, ChatGPT does not display a visible token count. It imposes message limits, which are stricter for modes requiring more computation.

Usage Limits of ChatGPT Based on the Subscription Plan

Message caps vary according to the subscription:

• Free: Up to 10 messages with GPT-5.5 every 5 hours, then switching to a lighter version.
• Go and Plus: Up to 160 messages with GPT-5.5 every 3 hours.
• Plus and Business: GPT-5.5 Thinking in manual selection is limited to 3,000 messages per week.

It is important to note that the automatic switch from Instant to Thinking does not count towards this weekly cap in manual selection. Manual choices thus allow for better management of limits. For Enterprise and Edu offers, where pricing is flexible, the adjustment directly influences the consumption of credits specific to each model.