OpenAI Drops the 5-Hour Cap on Codex, and Anthropic Flinches Again

Hello everyone, I'm Kakarot.

I saw some news this morning: Claude has once again extended the usage period for Fable 5 in its subscription plan to July 19th.

And this is already the third extension 😁.

Anthropic is really scared by OpenAI.

Last week, the release of the GPT 5.6 Sol model snatched away a large number of Claude model users.

Moreover, compared to the Fable 5 model, GPT 5.6 is extremely cheap, so ordinary consumers will definitely vote with their feet.

To quote Musk, a large number of ordinary users don't even use Fable 5-level models.

Then I saw another post.

OpenAI's Product Lead Tibo announced the temporary removal of the 5-hour limit on Plus, Business, and Pro plans.

This means you can go crazy building things! Only the weekly quota limit remains.

Then I opened my Codex, and wow, the 5-hour limit is really gone!

(If you're going to do this, I'm not going to sleep!)

Tibo is truly the strictest father to Anthropic!

You can see that Tibo posted this just 1 hour after Claude announced the reset.

Without a doubt, this is a direct attack on Claude!

Tibo also mentioned that ChatGPT's active users have reached 6 million.

That's terrifying, and a large part of it is likely due to Codex. (Of course, it's now been renamed to ChatGPT 🤔)

Honestly, these past two weeks, the major AI companies abroad have been fighting too fiercely.

xAI's Grok 4.5 made a comeback, with its model capability entering the first tier. OpenAI released the stunning Sol model, forcing Fable 5 to be extended three times in a row...

We seem to have forgotten a name. Google.

Google, which created the Transformer, holds TPUs, and possesses the most data in the world.

In this AI battle royale, it seems to have fallen out of the pack.

According to rumors, Gemini 3.5 Pro was supposed to be released long ago. It was held back by Fable 5 and GPT 5.6 Sol. Because launching it now, honestly, would just make it a sideshow. They lack confidence in their own model!

Or rather, Fable 5 and GPT 5.6 are just too strong!

For us ordinary folks, we don't care who's fighting whom. Models getting stronger, prices getting lower, and restrictions getting fewer — these are tangible benefits.

This is the direct benefit brought by competition 🤔

So, let's look forward to our domestic large models. GLM, Kimi, Deepseek, Hunyuan...

With the 5-hour limit removed, let's go crazy building with Codex, friends.

I'm Kakarot, continuously sharing useful AI information with you!