Modelpublic

mudler/Qwopus3.6-35B-A3B-v1-APEX-GGUF

sourceHugging Faceapache-2.0updated 5mo agoView on Hugging Face

24likes2.1kdownloads

Model Card

Qwopus 3.6 35B-A3B v1 APEX GGUF

APEX (Adaptive Precision for EXpert Models) quantizations of Jackrong/Qwopus3.6-35B-A3B-v1.

Brought to you by the [LocalAI](https://github.com/mudler/LocalAI) team | APEX Project

Available Files

File	Profile	Size	Best For
Qwopus3.6-35B-A3B-v1-APEX-I-Quality.gguf	I-Quality	23 GB	Highest quality with imatrix
Qwopus3.6-35B-A3B-v1-APEX-Quality.gguf	Quality	23 GB	Highest quality standard
Qwopus3.6-35B-A3B-v1-APEX-I-Balanced.gguf	I-Balanced	25 GB	Best overall quality/size ratio
Qwopus3.6-35B-A3B-v1-APEX-Balanced.gguf	Balanced	25 GB	General purpose
Qwopus3.6-35B-A3B-v1-APEX-I-Compact.gguf	I-Compact	17 GB	Consumer GPUs, best quality/size
Qwopus3.6-35B-A3B-v1-APEX-Compact.gguf	Compact	17 GB	Consumer GPUs
Qwopus3.6-35B-A3B-v1-APEX-I-Mini.gguf	I-Mini	14 GB	Smallest viable, fastest inference

What is APEX?

APEX is a quantization strategy for Mixture-of-Experts (MoE) models. It classifies tensors by role (routed expert, shared expert, attention) and applies a layer-wise precision gradient — edge layers get higher precision, middle layers get more aggressive compression. I-variants use diverse imatrix calibration (chat, code, reasoning, tool-calling, agentic traces, Wikipedia).

See the APEX project for full details.

Architecture

—Base Model: Jackrong/Qwopus3.6-35B-A3B-v1
—Architecture: Qwen3.5-MoE 35B-A3B
—Layers: 40
—Experts: 256 routed (8 active per token)
—Total Parameters: ~35B
—Active Parameters: ~3B per token
—APEX Config: 6+6 symmetric edge gradient across 40 layers
—Calibration: v1.3 diverse dataset (chat, code, reasoning, tool-calling, multilingual)

Run with LocalAI

bash

local-ai run mudler/Qwopus3.6-35B-A3B-v1-APEX-GGUF@Qwopus3.6-35B-A3B-v1-APEX-I-Balanced.gguf

Credits

APEX is brought to you by the LocalAI team. Developed through human-driven, AI-assisted research. Built on llama.cpp.