Downloads · 30 days
219
11% of all-time downloads
claymorecrystal/gemma-4-26B-A4B-it-heretic-APEX-GGUF
gemma-4-26B-A4B-it-heretic-APEX-GGUF is a machine learning model from claymorecrystal. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as gemma.
<div style="background-color: f59e0b; color: white; padding: 20px; border-radius: 10px; text-align: center; margin: 20px 0;" <h2 style="color: white; margin: 0 0 10px 0;"⚡ Each donation = another big MoE quantized</h2…
Downloads · 30 days
219
11% of all-time downloads
All-time downloads
2K
Public
Repo size
125 GB
Likes
1
Public
Click a slice to open those files.
.gguf125 GB · 100%
From the Hugging Face model README
APEX (Adaptive Precision for EXpert Models) quantizations of gemma-4-26B-A4B-it-heretic — an abliterated (uncensored) version of Gemma 4, created with the Heretic tool (v1.2.0) using Arbitrary-Rank Ablation (ARA) on layers 10-30 to reduce refusals while preserving capabilities (KL divergence 0.0499 from original).
Brought to you by the LocalAI team | APEX Project | Technical Report
Benchmarks coming soon. For reference APEX benchmarks on the Qwen3.5-35B-A3B architecture, see mudler/Qwen3.5-35B-A3B-APEX-GGUF.
| File | Profile | Size | Best For |
|---|---|---|---|
| gemma-4-26B-A4B-heretic-APEX-I-Balanced.gguf | I-Balanced | ~19 GB | Best overall quality/size ratio |
| gemma-4-26B-A4B-heretic-APEX-I-Quality.gguf | I-Quality | ~20 GB | Highest quality with imatrix |
| gemma-4-26B-A4B-heretic-APEX-Quality.gguf | Quality | ~20 GB | Highest quality standard |
| gemma-4-26B-A4B-heretic-APEX-Balanced.gguf | Balanced | ~19 GB | General purpose |
| gemma-4-26B-A4B-heretic-APEX-I-Compact.gguf | I-Compact | ~15 GB | Consumer GPUs, best quality/size |
| gemma-4-26B-A4B-heretic-APEX-Compact.gguf | Compact | ~15 GB | Consumer GPUs |
| gemma-4-26B-A4B-heretic-APEX-I-Mini.gguf | I-Mini | ~13 GB | Smallest viable, fastest inference |
| mmproj.gguf | Vision projector | ~1.2 GB | Required for image understanding |
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, technical report, and scripts.
local-ai run mudler/gemma-4-26B-A4B-it-heretic-APEX-GGUF@gemma-4-26B-A4B-heretic-APEX-I-Balanced.gguf
APEX is brought to you by the LocalAI team. Developed through human-driven, AI-assisted research. Built on llama.cpp.