How to Maximize the New Unlimited Free ChatGPT 23x More Effectively

"Free ChatGPT will just stop working soon, right?"

That premise was shattered on August 6th. ChatGPT's free plan is becoming unlimited.

However, if you keep using it the same way you do now, the benefit will be almost zero. Those who haven't broken the habits from the era of usage limits will end up with nothing changed.

Do any of these sound familiar?

• You spend minutes refining a prompt because you're afraid of hitting the limit.
• You've been stopped at a crucial moment and had to wait hours to continue.
• You pay for a subscription but mostly only use it for text.
• You've decided the free version is useless and don't even open it.

If even one of these applies, please read on. First, I'll organize what has changed and to what extent. Then, I've prepared prompts you can use starting today.

Finally, I've included criteria for deciding whether to continue paying, so please bookmark this and try them one by one.

As a quick announcement,

I will be holding an AI × 𝕏 Zero-to-One Hundred Study Session for 5 days starting August 10th. I, who achieved 100 million yen in annual sales as a side hustle, will explain everything from how to generate hundreds of thousands of impressions even with a 2-digit follower count to how to connect that to monetization. Please make a reservation here. ↓

https://utage-system.com/p/nzhojCEUqoAQ?ftid=KtFj0pUrCa5I

Now, let's get to the main topic.

What Changed on August 6th

There are three main changes.

1. Model Change

The first is that the default model for the Free and Go plans has changed to GPT-5.6 Luna. This replaces the previously running GPT-5.5 Instant.

The model is the engine that runs ChatGPT.

Even if the screen looks the same, if the engine is replaced, the accuracy of the answers changes. The core of the ChatGPT used by free users has been completely renewed.

According to OpenAI's announcement, answers containing factual errors have decreased by about 62% compared to GPT-5.5 Instant. However, since these are OpenAI's own internal figures, take them with a grain of salt.

2. Unlimited Text Chat

The second change is the big one.

Text chat is becoming unlimited. The rollout starts the week of August 10th. If you open it now and it hasn't changed yet, nothing is broken; it's just coming soon.

Previously, limits often kicked in mid-chat, making it unusable, but that's gone now, so you can use it to your heart's content without worrying about restrictions.

3. Addition of the Think Feature

A "Think" button has also been added in this update. When pressed, it takes more time to think before answering.

This is actually a major update.

The previous GPT-5.5 Instant was fast but lacked a reasoning mode.

For difficult problems, having it spend time reasoning significantly improves the accuracy of the response, so please give it a try.

However, Unlimited is for Text Only

Let me mention some precautions first. Only text interactions have become unlimited.

• File uploads
• Image generation
• Other tools

Limits remain for these as before.

Also, even though it's called "unlimited," anti-abuse limits still apply. Usage that involves mechanical, rapid-fire posting may be stopped.

If you go in thinking everything is free and unlimited, you'll definitely get stuck with image generation. Text generation is unlimited, but everything else remains the same.

Please keep this in mind.

The Correct Way to Use the Free Plan Has Changed

This is the core of the matter.

Until now, the "correct" way was to pack everything into one long, perfect instruction.

That was right because usage limits were precious.

That premise is gone. From now on, the correct way is to refine the output while showing the progress.

First, send a short prompt.

Look at what comes out.

Reply by saying, "This part is weak, so fix it."

Stop trying to hit the mark in one shot and use it as a partner working beside you.

To be honest, since Luna is a lightweight model, the quality of a single answer won't reach that of the paid high-end models.

But if it's a finished product after multiple rounds of drafts and corrections, that's a different story. The quality of the final result isn't determined solely by the model's intelligence. It's determined by how much human refinement was involved.

Previously, this process of adding manual touches to improve quality was impossible due to usage limits, but now it can be done even on the free plan.

So, you no longer need to spend time crafting long prompts. Start by using the prompts in the next chapter as they are.

Usage You Can Start Today

I've written down ways you can use it immediately, so please rewrite them with your own work vocabulary.

1. Have it Inspect the Answer

Step 1: Ask your question as usual.

Step 2: Send this in response to the answer.

1Inspect the current response as a harsh editor.
2Point out weak points from the following three perspectives:
3- Parts where facts are suspicious
4- Parts where logic is jumping
5- Parts where specificity is lacking
6You don't need to praise me. Explain why each part is weak.

Step 3: Based on the feedback, decide which points to fix and reply like this:

1Provide the full text of a revised version incorporating the parts I've selected.
2Do not change the parts that weren't pointed out.
3Finally, tell me what you fixed and how.

Repeat Step 2 and Step 3 two or three times. When the feedback starts becoming about minor phrasing, that's the time to stop.

You should always run this once before showing text to others. It's faster to have the AI point out weaknesses than to re-read it yourself.

The reason surprisingly few people do this is simple: on the free plan with usage limits, they couldn't afford to spend multiple chat turns on a single answer.

From now on, since it's unlimited, you should let ChatGPT handle the checking and proofreading through as many cycles as needed.

2. Get 5 Ideas and Take the Best Parts

So many people stop after getting one idea. It's a waste.

Step 1: Send this.

1Provide 5 ideas for this theme, each with a completely different angle.
2Conventional ideas are forbidden. Include one curveball that you wouldn't normally choose.

Step 2: Choose two numbers you liked and send this.

1Combine the good parts of #1 and #4 into a single version.
2Don't just add them together. Think of a way where the strengths of both stand out simultaneously.
3Write down the elements you discarded during the synthesis at the end.

Step 3: The finishing touch is this.

1Make it sharper.
2Remove all bland expressions.
3State things clearly where possible. Cut out generalities that apply to anyone.

With these steps, it becomes something entirely different from the first single idea. By comparing five options, you understand what you actually think is good. You can't see that if you only look at one.

Should You Keep Paying or Go Back to Free?

Please apply the following questions to yourself.

If any of these four apply to you, it's still worth paying even with the unlimited update:

• You frequently use image generation or file analysis.
• Quality of the answer is more important than the number of uses.
• You want to be on the side of creating Custom GPTs.
• You also use things like Codex.

Summary

• The standard model for Free and Go has changed to GPT-5.6 Luna. Text chat becomes unlimited starting the week of August 10th.
• Text chat usage is now unlimited. However, limits remain for images and files as before.
• Stop trying to hit the mark in one shot with long instructions. Send short prompts and layer corrections.

Since the range of what can be done on the free plan has increased, I hope you take this opportunity to use ChatGPT to its fullest. If you find value in it, then consider paying. ChatGPT and Codex can definitely pay for themselves if used properly, so please make the most of them.

Finally, I usually post about the latest AI information and AI-based monetization methods. If this was helpful, I'd be happy if you followed me.

Also, I will be holding an AI × 𝕏 Zero-to-One Hundred Study Session for 5 days starting August 10th.

I, who achieved 100 million yen in annual sales as a side hustle, will explain everything from how to generate hundreds of thousands of impressions even with a 2-digit follower count to how to connect that to monetization. Please make a reservation here.

↓

https://utage-system.com/p/nzhojCEUqoAQ?ftid=KtFj0pUrCa5I

Furthermore, I am currently giving away a set of 55 major benefits, including textbooks and implementation methods for Claude Code and Codex, as well as AI-based monetization methods, for free. You can receive them immediately from the link below, so if you haven't received them yet, please do so.

↓

https://utage-system.com/line/open/cwgwX1a35XDK?mtid=FNAamIuYaEet