he was released about a month ago, and after three days, he was taken out of the public for security reasons. However, it has now been restored to Claude's subscription, and anyone with a Claude subscription can access Claude Fable 5.
Unfortunately, Anthropic has limited the amount of usage you get with Claude Fable 5 before reaching usage limits, limiting you to 50% of your weekly limits. Therefore, you cannot simply use Claude Fable in all your activities, because you will quickly reach the limits of use and you will not be able to use it anymore.
In this article, I'll cover how you can get the most out of Claude Fable 5, while highlighting how you can use it to edit, edit, and update code instead of doing the boring work of running some code.
I'll discuss how you can get Fable level intelligence but stay within the limits of Fable. These are the strategies I use every day to get the most out of my Claude Code subscription.
Why use Claude Fable 5
The main reason you should use Claude Fable 5 is that it is the most powerful coding model yet.
I've tried basically all the major models for writing labs, including:
- Google Gemini
- GLM 5.2
- The OpenAI Codex
OpenAI Codex is the biggest competitor, and I would say that their models are very comparable to Anthropic's second best model, Claude Opus 4.8. I would say that in most cases GPT-5.5 and definitely GPT-5.6 is now better than Claude Opus 4.8, but not better than Claude Fable.
There are a few things I can highlight that Claude Fable is much better than any other model:
- Finds problems with the code
- Finding opportunities to rework and improve separation of concerns, do not repeat, and other coding principles.
- Planning by looking at the repo, launching research agents, and making a good plan for how to successfully deploy the feature
These are the things Claude Fable is best at. However, you should note that I did not directly mention the implementation of the code because I believe that Claude Fable is not so high in the implementation of the code compared to Claude Opus 4.8, for example, which I will cover in the next section.
How to expand Claude Fable 5
My code pipeline
The main way to get the most out of Claude Fable 5 without using all the usage limits is to not use Fable for pure code use, but basically use it for everything else.
At a high level, my coding pipeline works like this:
- Use Fable to schedule implementation or fix bugs
- Use Claude Opus 4.8 or GPT-5.6 to run the code
- Use GPT-5.6 to update the code
- Compile the code in dev
The good thing about this is that you only use Fable for editing, which is also where the biggest difference between Fable and other models is.
So you don't use Fable tokens to run the code, Claude Opus 4.8 or GPT-5.6 work equally well for it. This saves a lot of usage restrictions in Claude Fable, which you can use to schedule a lot of tasks.
In some cases, you can also use Claude Fable to update the code. However, in my experience, Codex is good enough at updating the code itself, so you don't need to use Fable for it.
In addition, Claude Opus 4.8 is my main driver for running code, and I like to use a different model to review code, making Codex the best alternative.
How to plan successfully with Claude Fable
In my last section, I covered my top code pipeline so I don't waste tokens on Claude Fable when it's unnecessary, and you can complete the same task with Claude Opus 4.8.
However, another skill you need is how to plan successfully with Claude Fable. How do you find and create a clear plan that other agents can use? How do you look at opportunities for innovation? How do you tell it to find bugs and make sure it has all the tools accessible, etc.?
The main point I want to get here, compared to previous models, is that you have to make the Claude Fable work more autonomously than you think.
You can even start tasks that you are not sure how to do or what it would look like. Start chatting with Claude Fable and make the best solutions come up.
Usually what I do is explain the situation to Claude Fable. For example, I want to use a certain feature. Then I have Claude Fable researching the archive I'm currently in. Come up with different ideas about the architecture and logic of what the implementation could look like, and present the results to me in HTML. If it helps to use it for visuals. I tell Claude Fable to use as many images as possible for use.
Note, however, that I am asking Claude Fable to make that decision about whether the visuals can help or not, and it should make all the decisions about how to present them as easy as possible for me. I tried to instruct the model as little as possible, without highlighting the work I want and how to make sure the work is done correctly.
Something that Claude Fable does well is everything in between. So you provide the work, which must be done, to ensure how it is done, and Claude Fable can complete all the work in the middle, or at least make a plan of how to complete all the work, which you can use a model like Claude Opus 4.8 to really use.
How to remake Claude Fable
Refactoring is becoming more and more appropriate, and now we have dozens of coding agents that run almost every code.
I have discussed the topic of recycling a lot in my previous articles, which can be summarized in the following:
If code execution starts taking longer than expected or more bugs than before are introduced, it's usually a sign that the code needs to be refactored.
Therefore, you should keep an eye out for when things start taking longer than they should, which is a clear sign that you should start refactoring the code. Furthermore, I don't think the fact that you have to redo the code means you did something wrong before. It's just an evolution of working with coding agents. Once you have used enough new code, you should always re-enter it. And luckily, refactoring can be easily done with coding agents, so it doesn't take much effort and isn't what you're looking for in terms of understanding.
To redo, you can issue a command like:
Look for refactoring opportunities in this repository, what should be
improved, look for bad coding practices, don't repeat yourself, poor
separation of concerns and so onAnd this will definitely work. If you've never done a remake before, you should definitely do it right away and have Claude do all the recommended actions. However, there is a better way to recode with Claude Fable.
Basically, you as a person should detect the symptoms yourself, for example, by noticing that the execution of the code in a certain area is taking longer than expected.
Once you figure this out, you write down what area of the code this is happening in, and point Claude directly to this area. For example, you could say:
The last time we implemented code in the processing pipeline, it took a
lot longer than expected and I think this is a sign that we need to
improve the code quality in that area. Research the coding session I ran
there and look for refactoring opportunities so that we can increase speed
and reduce bugs in the futureThis works mainly for two main reasons.
- You point to Claude's Code somewhere where you know there are problems
- It provides Claude code logs from your coding agent where you can see what happened and why it took so long, giving you a better understanding of what needs to be redone.
You then use this message, tell Claude Code to give you the results in HTML, in order of importance, and then tell Claude Code to do all the optimization.
The conclusion
In this article, I have put together how to get the most out of Claude Fable 5. Unfortunately, we have limited usage limits, and with the Claude Fable model, you just have to use it effectively. In short, I use Claude Fable to program and use models like Claude Opus 4.8 or GPT-5.6 to actually run the code, saving the use of multiple tokens with Fable. In my opinion, this does not have a big impact on the quality of the code because usually running the code is easier than planning what to do, and the difference between Fable and other models is very small when it comes to the actual code execution. I believe we will see this more in the future, where you have incredibly powerful models available in limited quantities, which you have to use with caution due to cost or limitations. You just have to think about how you can use them as effectively as possible without wasting all the usage or accumulating a lot of expenses.
👋 Touch
👉 My Free eBook and Webinar:
🚀 10x Your Engineering with LLMs (3 Day Free Course)
📚 Get my free ebook Vision Language Models
💻 My webinar on Vision Language Models
👉 Find me on social media:
💌 Stack
🐦 X / Twitter
