Qwopus3.6-35B-A3B-v1 MLX bf16 (reference)

Full non-quantized MLX bfloat16 conversion of Jackrong/Qwopus3.6-35B-A3B-v1. Reference build for this MLX ladder.

Converted directly from the original HF bf16 safetensors. The 3/4/6/8-bit siblings are independently quantized from the same source — not chained from this repo, not converted from GGUF.

The full MLX ladder

Variant	Repo	Disk	~Min unified RAM	Role
MLX bf16 (this repo)	this	69.3 GB	~72 GB	Reference
MLX 8bit	`Qwopus3.6-35B-A3B-v1-MLX-8bit`	36.8 GB	~40 GB	Near-lossless
MLX 6bit	`Qwopus3.6-35B-A3B-v1-MLX-6bit`	28.2 GB	~32 GB	Quality / size middle
MLX 4bit	`Qwopus3.6-35B-A3B-v1-MLX-4bit`	19.5 GB	~22 GB	Standard daily-use tier
MLX 3bit	`Qwopus3.6-35B-A3B-v1-MLX-3bit`	15.2 GB	~18 GB	Smallest practical

Collection: Qwopus3.6-35B-A3B-v1 MLX

Use

pip install mlx-lm
mlx_lm.generate --model zaydiscold/Qwopus3.6-35B-A3B-v1-MLX-bf16 \
  --prompt "Explain quantum entanglement in one paragraph" --max-tokens 200

Conversion

python -m mlx_lm convert \
  --hf-path Jackrong/Qwopus3.6-35B-A3B-v1 \
  --mlx-path ./Qwopus3.6-35B-A3B-v1-MLX-bf16 \
  --dtype bfloat16

Notes

This repo is not quantized.
Use it as the reference for re-quantizing to other formats.

Credits

Source: Jackrong/Qwopus3.6-35B-A3B-v1
MLX conversion: zaydiscold

Downloads last month: 155

Safetensors

Model size

35B params

Tensor type

BF16

MLX

Hardware compatibility

Quantized

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for zaydiscold/Qwopus3.6-35B-A3B-v1-MLX-bf16

Base model

Qwen/Qwen3.6-35B-A3B

Finetuned

unsloth/Qwen3.6-35B-A3B

Adapter

Jackrong/Qwopus3.6-35B-A3B-v1

Finetuned

(2)

this model

Collection including zaydiscold/Qwopus3.6-35B-A3B-v1-MLX-bf16

Qwopus3.6-35B-A3B-v1 MLX

Collection

Complete MLX quantization grid for Qwopus3.6-35B-A3B-v1 — bf16/8/6/4/3-bit, every quant converted directly from HF bf16. None chained. • 5 items • Updated 26 days ago