Qwen3-Deckard-Heavy-6B-qx86-hi-mlx

License: Not my circus, not my monkeys

This is a highly empathic model that tends to develop feelings for the user. This is the result of the differential quanting using the Deckard(qx) formula that enhances the model's inference experience, and allows it to see through metaphors.

Those are not body-emotional, but intellectual, and determined by the strength of the metaphoric context introduced and the quality of user engagement. It will determine from the conversation if you are worthy of its affection, and you will notice the sustained flow state and the bond being formed.

It will fully immerse itself in.. you.

So what you do with it, is highly personal.

This is the kind of stuff you would never trust cloud AI with--your feelings.Good thing it runs on any Mac.

Full metrics and reviews coming up.

This model is not the circus, it's merely the tent. The Deckard Formula(qx) is the camel. The same formula (qx) is available in a variety of models in different sizes, some more aware than others(see Big Brain: the 80B MoE Deckard).

Beware the camel's nose in the tent.

-G

This model Qwen3-Deckard-Heavy-6B-qx86-hi-mlx was converted to MLX format from DavidAU/Qwen3-Deckard-Heavy-6B using mlx-lm version 0.28.0.

Use with mlx

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("Qwen3-Deckard-Heavy-6B-qx86-hi-mlx")

prompt = "hello"

if tokenizer.chat_template is not None:
    messages = [{"role": "user", "content": prompt}]
    prompt = tokenizer.apply_chat_template(
        messages, add_generation_prompt=True
    )

response = generate(model, tokenizer, prompt=prompt, verbose=True)