          arc   arc/e boolq hswag obkqa piqa  wino
qx86-hi   0.635,0.821,0.891

Quant     Perplexity      Peak Memory   Tokens/sec
qx86-hi   4.489 ± 0.029   45.50 GB      1448

Baseline model

Qwen3.6-35B-A3B

brainwaves

         arc   arc/e boolq hswag obkqa piqa  wino
mxfp8    0.581,0.757,0.892,0.751,0.428,0.803,0.688
qx86-hi  0.576,0.742,0.896,0.745,0.422,0.803,0.708
qx64-hi  0.589,0.763,0.892,0.748,0.428,0.801,0.700
mxfp4    0.586,0.767,0.886,0.751,0.428,0.798,0.681

Thinking toggle

This model quant is using(an early version of) the fixed jinja template from froggeric/Qwen-Fixed-Chat-Templates

Drop <|think_on|> or <|think_off|> anywhere in your system or user prompt. The template intercepts the tag, removes it from context so the model never sees it, and flips the mode.

I added: <|think_forget|> / <|think_remember|> if you want to exclude thinking content out of the context (experimental).

More metrics coming soon.

-G

Use with mlx

pip install mlx-lm

python

from mlx_lm import load, generate

model, tokenizer = load("Qwen3.6-35B-A3B-Fable-5-Distill-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, return_dict=False,
    )

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

nightmedia/Qwen3.6-35B-A3B-Fable-5-Distill-qx86-hi-mlx

Qwen3.6-35B-A3B-Fable-5-Distill-qx86-hi-mlx

Baseline model

Thinking toggle

Use with mlx

nightmedia/
Qwen3.6-35B-A3B-Fable-5-Distill-qx86-hi-mlx