Should your AI assistant always run at full throttle, even for a simple rephrasing? It’s a bit like using a Formula 1 car to go grocery shopping. OpenAI, the company behind ChatGPT, has just integrated a discreet but essential feature: manual control over the algorithm’s effort level, message by message. A simple gesture that changes the game for your daily usage and limit management.
This new option, integrated directly into the mobile app, allows you to choose between three ‘thinking’ modes: Instant, Thinking, and Extended. The idea is to adapt the artificial intelligence’s computational power to the actual complexity of your query. No more wasted resources for basic tasks; instead, embrace a more agile and conscious use of the tool.
Beyond mere novelty, this approach lays the groundwork for a more nuanced interaction with AI systems. It encourages us to better understand how these models process information and to become true conductors of their performance. This is a key skill to develop to get the most out of these technologies.
Unleash or Restrain AI: A New User Control
The ‘thinking’ level you assign to ChatGPT determines the intensity of effort the model dedicates to your message. The higher this level, the more time the algorithm takes to reason. Response quality then improves for complex problems, but wait times increase, and crucially, your usage limits deplete faster. Conversely, a low level guarantees an almost instant response while consuming fewer resources.
This distinction highlights a fundamental technical reality: artificial intelligence is not a monolithic entity. Its capabilities can be modulated, and this modulation has direct consequences on performance and consumption. This is an opportunity for users to master their AI consumption, much like managing a smartphone’s battery.
Enter your message
Start by writing your query or instruction in the text field of the ChatGPT mobile app.
Long-press "Send"
Instead of briefly tapping the send arrow, hold your finger on it for a few moments.
Choose your level
A selector will appear. Select the desired level (Instant, Thinking, or Extended), then confirm sending.
Managing Your AI Credits: The Art of Digital Frugality
The main reason for this setting? To preserve your valuable message limits. Unlike some competitors like Claude, which display a visible token count, ChatGPT operates with message caps, which are stricter for intensive reasoning modes. These modes, by mobilizing more computational power, consume your quota faster.
Message economy is therefore at the heart of this feature. If you frequently use ChatGPT for varied tasks, knowing how to calibrate the AI’s effort becomes an essential strategy. It’s a way to extend your subscription or avoid switching to a lighter version too quickly.
For Enterprise and Edu plans, where pricing is flexible and based on credit consumption, the setting has an even more direct impact on the wallet. Each interaction in ‘Thinking’ or ‘Extended’ mode consumes more credits. Choosing wisely also means managing your AI budget.
AI EFFORT MANAGEMENT: CHATGPT VS CLAUDE
Granular control per message
Effort level set globally
Direct limit/cost optimization
So, ready to become the conductor of your AI’s ‘thinking’? This small ChatGPT adjustment reminds us of one essential thing: the most effective artificial intelligence isn’t the one that always does the most, but the one you know how to use just enough. How will this new granularity redefine our productivity and our relationship with intelligent assistants?
Chargement de la galerie…
About Rigaud Mickaël
Passionate about tech and a Linux enthusiast, I decipher AI with a unique and intense vision to make it useful to all, between robots, rock and the geek universe.
