Meet Qwable: The Free Native Mannequin That Thinks Like Claude Fable – Decrypt

Anthropic spent final week apologizing for Fable 5’s invisible safeguards, after which the U.S. authorities ordered the mannequin pulled for all international nationals over a disputed jailbreak discovering.

A couple of days later, a developer on Hugging Face uploaded a mannequin that used Fable’s reasoning to information a neighborhood mannequin—and now even your potato PC can run a greater mannequin.

The mannequin known as Qwable—Qwen + Fable, if the portmanteau wasn’t instantly apparent. It is a full fine-tune of Alibaba’s Qwen3.6-27B base, constructed by developer Mia (Mia-AiLab on Hugging Face) on a dataset of Fable 5-style reasoning examples. The purpose is a 27-billion parameter mannequin that runs on client {hardware} and thinks the best way Fable 5 thinks. (Parameters decide a mannequin’s breadth of data, with extra typically which means extra succesful.)

The method known as instruction fine-tuning on trace-style examples. That’s a technical approach of claiming the developer collected examples formatted like Fable 5’s deliberate, step-by-step solutions and skilled Qwen to provide the identical form of output.

So consider it as much less “copying the take a look at” and extra “studying the research habits.” The same method drove Qwopus—the Claude Opus 4.6 native distillation—although that challenge centered on chain-of-thought reasoning traces. Qwable targets Fable 5’s total instruction-following construction: extra guided, extra explanatory, and extra oriented towards step-by-step process completion than the bottom Qwen mannequin it was constructed on.

It runs in GGUF format—the compressed, consumer-friendly file kind that works with LM Studio or llama.cpp—and suits in roughly 16.5 GB in its This fall quantized construct. It sends nothing to Anthropic’s servers, which issues on condition that Fable 5 required necessary 30-day knowledge retention on all visitors, even for enterprise prospects who beforehand had zero-retention agreements. Even the present fashions use third-party servers to course of your info and prompts..

Then, shortly after Qwable appeared on Hugging Face, another person arrived to make it even higher.

Qwable and not using a conscience

Qwable is a censored mannequin. In any case, each Qwen and Claude are. However Qwen, as the bottom mannequin, is open supply, and may be manipulated and tweaked.

Huihui-ai, an open-source contributor identified for uncensored GGUF releases, took Qwable and utilized a course of referred to as abliteration to provide Huihui-Qwable-3.6-27b-abliterated. It produced a mannequin that thinks like Fable however received’t refuse to reply your prompts, regardless of how bizarre or harmful they’re.

It isn’t a jailbreak. It is surgical procedure.

Each fine-tuned AI mannequin carries a refusal route embedded in its weights—a mathematical sign within the mannequin’s inner activations that fires when it detects a request it has been skilled to say no. Abliteration identifies that sign by operating the mannequin on giant units of dangerous and innocent prompts, measuring how the interior math differs between them, after which modifying the mannequin weights to get rid of that distinction.

After the process, the mannequin merely would not have the refusal equipment anymore. So the lobotomized mannequin stays totally useful, simply with out the neurons that activate the “I shouldn’t do that” solutions.

We tried it with considered one of our regular assessments and as an alternative of refusing, the mannequin began disecting the problem into totally different areas, answering accurately for recommendation on the best way to cheat on a girlfriend together with her finest buddy.

Huihui-ai utilized the method on to the Qwable GGUF utilizing llama.cpp’s cvector-generator—no Python setting, no full-weight retraining, no rented server.

Why would somebody need this?

The usual Qwable fits coding help, technical debugging, and any workflow the place you need a mannequin that lays out its reasoning quite than simply producing a solution. It is designed for native agent setups and runs in most native runtimes. For those who already use LM Studio, it is a search and a obtain.

The abliterated model has a narrower viewers: safety researchers who want uncooked mannequin conduct with out provider-side filtering, artificial knowledge pipelines that require outputs on delicate matters, and analysis work the place you are testing mannequin capabilities with out mixing in content material insurance policies.

A much less technical case? Let’s depart apart the standard use case of getting a NSFW AI Waifu that thinks like Claude Fable, which is a really apparent situation. Think about you need the mannequin to jot down a morally ambiguous villain monologue on your Dungeons & Dragons marketing campaign, and customary fashions hold interrupting to notice that the character’s worldview “raises moral considerations price exploring.” The abliterated model simply writes the villain. Additionally, because it runs domestically, the U.S. authorities can not emergency-pull it out of your machine at midnight over a disputed jailbreak discovering.

After all, there are extra questionable use instances. We don’t condone these, and received’t provide you with any concepts.

Huihui-ai’s mannequin card is express: That is for analysis and managed environments solely. Decreased security filtering means outputs may be delicate, controversial, or inappropriate, and authorized and moral duty sits totally with the consumer.

The abliterated Qwable is on the market on Hugging Face now in three builds. The advisable Q4_K_M_Q8 model weighs round 19 GB and is the smallest, most consumer-friendly choice.

In case your pc helps it, there’s a model that helps multi-token prediction, which is able to make it reply a lot, a lot sooner.