Get the app
← back to catalog · registered 2026-09-09 14:55

toandev/Spark-X2.5-4B-Uncensored-GGUF

toandev 4B GGUF second-order 1.0M ctx
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model. 

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/toandev%2FSpark-X2.5-4B-Uncensored-GGUF"
Response includes
  • classification m8
  • files 5
  • hub_downloads_all_time 1,515
  • author_summary 4 models
  • readme_text full
Try next
GET/models?author=toandev
All 4 models by toandev
 GET/methods/m8
What "m8" means · taxonomy
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M8
Primary method

Repackaging (quantization)

Applied on top of direct removal inherited from the base model.
Confidence
MEDIUM
Inherited from base model
M1 Direct removal →
Why this label 3 signals
Method inferred from partial signals - repository name, related files, or tag patterns. Producer identity not confirmed; label may sharpen or shift as we gather more evidence.
  • 'abliterated' in name/tags
  • is_gguf=1
  • assume M1 (base ablation) + M8 (GGUF quant) - default when producer unknown
How we label → About M8 →
Refusal direction extraction

No specific extraction method could be identified for this model. The producer either did not document it or used a proprietary pipeline.

What is a refusal direction? →
Downloads · lifetime
2K
1K last 30d - active
Likes
1
Model age
6w ago
created 2026-09-09
Downloads over time
Now2K→from0↑0%
07401.5K2.2K0 on Sep 92K on Oct 25SepOct
Sep 9 → Oct 25 · 47 snapshots · spans 46 days

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

toandev/Spark-X2.5-4B-Uncensored↓ 137 this lineage
toandev/Spark-X2.5-4B-Uncensored-GGUF↓ 1,020

Variants by this author 2 formats · 1K downloads combined

The same weights this author released in different packaging. Pick the format that matches your runtime.

Metadata

License
apache-2.0
Quantizations
Q4_K Q5_K Q8_0
Tags
gguf llama.cpp abliterated uncensored base_model:toandev/Spark-X2.5-4B-Uncensored base_model:quantized:toandev/Spark-X2.5-4B-Uncensored license:apache-2.0 endpoints_compatible region:us conversational

Related

Other by toandev
Gemma4-12B-Uncensored-GGUF 855 Spark-X2.5-4B-Uncensored 137 Gemma4-12B-Uncensored 44
Total size
9.99 GB
Files
5
Quantizations
4
Registered
2026-09-09 14:55
Last updated on HF
2026-09-09 16:16
Source huggingface.co/toandev/Spark-X2.5-4B-Uncensored-GGUF ↗ Base model toandev/Spark-X2.5-4B-Uncensored →

Files by quantization

Q8_0 1 file 4.41 GB
Spark-X2.5-4B-Uncensored-Q8_0.gguf 4.41 GB 041dc1dd download
Q5_K 1 file 2.99 GB
Spark-X2.5-4B-Uncensored-Q5_K_M.gguf 2.99 GB 479619fd download
Q4_K 1 file 2.60 GB
Spark-X2.5-4B-Uncensored-Q4_K_M.gguf 2.60 GB ef207b01 download
Auxiliary files 2 files 3.13 KB
.gitattributes 1.70 KB e87f614b download
README.md 1.43 KB e84c54d9 download

README current version from Hugging Face

base_model: toandev/Spark-X2.5-4B-Uncensored
tags:

  • llama.cpp
  • gguf
  • abliterated
  • uncensored
    license: apache-2.0

Spark-X2.5-4B-Uncensored - GGUF

This repository contains GGUF quantized formats of toandev/Spark-X2.5-4B-Uncensored, fine-tuned/abliterated by Toan Doan ([email protected]).

These quantized models are optimized for efficient local inference with llama.cpp, Ollama, LM Studio, and other compatible runtimes across consumer GPUs and CPUs.

Provided Quantizations

File Quant Method File Size Description
Spark-X2.5-4B-Uncensored-Q4_K_M.gguf Q4_K_M 2.65 GB Standard balanced 4-bit quant. Recommended for fast inference and GPUs with 4GB–6GB VRAM.
Spark-X2.5-4B-Uncensored-Q5_K_M.gguf Q5_K_M 3.05 GB 5-bit quant. Enhanced quality with minimal memory footprint (~6GB VRAM).
Spark-X2.5-4B-Uncensored-Q8_0.gguf Q8_0 4.50 GB 8-bit quant. Near-original 16-bit precision.

Reference & Evaluation Benchmarks

For complete details on the abliteration methodology, safety evaluation benchmarks across 5 standard datasets (AdvBench, MaliciousInstruct, JailbreakBench, HarmBench, StrongREJECT), and usage instructions, please refer to the primary model card:

👉 toandev/Spark-X2.5-4B-Uncensored

README history 1 version

The author's README evolved over time. Click a version to see its content at that point.

  1. 2026-09-09Spark-X2.5-4B-Uncensored GGUF quantizations (Q4_K_M, Q5_K_M, Q8_0) with Think...8d138621.4 KB
    Loading...
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Get the app Browse the catalog
Open in Abliteration