GGUF models by ngxson

37 GGUF models quantized by ngxson on Hugging Face. Sizes run from 1M to 673B parameters. The longest trained context is 10240K tokens.

Most downloadedNewestLargestLongest contextSmallest
TI
tinyllama split test
Tinyllama · ngxson
Params24M
Context—
Smallest0.0 GB
GGUF
TE
test gguf models
Test · ngxson
Params2B
Context32K
Q4_K_M1.2 GB
GGUF
DE
DeepSeek R1 Distill Qwen 7B abliterated
DeepSeek · ngxson
Params7.6B
Context128K
Q4_K_M4.7 GB
Reasoning
QW
Qwen2.5 7B Instruct 1M
Qwen · ngxson
Params7.62B
Context986K
Q4_K_M4.7 GB
Tool use
GL
GLM 4.7 Flash
GLM · ngxson
Params29.9B
Context198K
Q4_K_M18.1 GB
ReasoningTool use
SM
SmolLM2 1.7B Instruct Q4 K M
SmolLM · ngxson
Params1.71B
Context8K
Q4_K_M1.1 GB
GGUF
TE
test gguf lora adapter
Test · ngxson
Params1M
Context64
Smallest0.0 GB
GGUF
BO
boring testing tiny
Boring · ngxson
Params—
Context128
Smallest0.0 GB
GGUF
VI
Vintern 1B v3 5
Vintern · ngxson
Params630M
Context32K
Smallest0.7 GB
Tool useVision
MI
Home Cook Mistral Small Omni 24B 2507
Mistral · ngxson
Params23.6B
Context128K
Q4_K_M14.3 GB
Tool useVision
MI
Devstral Small Vision 2505
Mistral · ngxson
Params23.6B
Context128K
Q4_K_M14.3 GB
CodingVision
VI
Vistral 7B ChatML
Vistral · ngxson
Params7.29B
Context32K
Q4_K_M4.4 GB
GGUF
WL
wllama split models
Wllama · ngxson
Params2.61B
Context2K
Q4_K_M0.7 GB
GGUF
LL
Llama 3 Instruct Abliteration LoRA 8B
Llama · ngxson
Params88M
Context—
Smallest0.4 GB
GGUF
MI
MiMo VL 7B RL
Mimo · ngxson
Params7.62B
Context125K
Q4_K_M4.7 GB
Tool useVision
QW
qwen3 next tiny test
Qwen · ngxson
Params425M
Context8K
Smallest0.9 GB
Tool use
LF
LFM2 VL 450M
Lfm · ngxson
Params354M
Context125K
Smallest0.2 GB
Vision
GL
GLM 4.7 Flash small test
GLM · ngxson
Params3.02B
Context198K
Smallest3.2 GB
ReasoningTool use
QW
qwen3 next fixed
Qwen · ngxson
Params79.7B
Context256K
Smallest29.1 GB
Tool use
GE
Gemma 3 4b Pt
Gemma · ngxson
Params3.88B
Context128K
Smallest2.4 GB
Vision
MI
Mistral Large 3 675B Instruct 2512
Mistral · ngxson
Params673B
Context288K
Q4_K_M405.8 GB
Tool useVision
SM
SmolLM2 135M Instruct
SmolLM · ngxson
Params135M
Context8K
Smallest0.1 GB
GGUF
VI
vistral meow
Vistral · ngxson
Params7.29B
Context32K
Smallest4.4 GB
GGUF
LL
Llama 3.2 1B Creative Lora
Llama · ngxson
Params11M
Context—
Smallest0.0 GB
GGUF
LL
Meta Llama 3.1 8B Instruct
Llama · ngxson
Params8.03B
Context128K
Q4_K_M4.9 GB
Tool use
LL
Llama 4 Scout 17B 16E Instruct
Llama · ngxson
Params108B
Context10240K
Smallest114.5 GB
Tool use
GL
glm4.5 small test
GLM · ngxson
Params22.8B
Context128K
Q4_K_M13.4 GB
ReasoningTool use
LL
MiniThinky v2 1B Llama 3.2 Q8 0
Llama · ngxson
Params1.24B
Context128K
Smallest1.3 GB
GGUF
TE
test model preset
Test · ngxson
Params81M
Context125K
Q4_K_M0.1 GB
Tool use
LL
LoRA Hermes 3 Llama 3.1 8B
Llama · ngxson
Params185M
Context—
Smallest0.7 GB
GGUF

About ngxson on Hugging Face

ngxson has published 37 GGUF models that ggufmodels.com tracks, across families such as Llama, Test and Qwen. The most downloaded are tinyllama split test, test gguf models and DeepSeek R1 Distill Qwen 7B abliterated.

The quants ngxson publishes most often are Q4_K_M (14), Q8_0 (10), F16 (8) and Q2_K (3). Their Q4_K_M files range from 0.1 to 405.8 GB. Parameter counts run from 1M to 673B, and the longest trained context window is 10240K tokens.

Frequently asked questions

What GGUF models has ngxson published?

37 models are listed here, mainly from the Llama, Test and Qwen families. The most downloaded is tinyllama split test.

Which quants does ngxson publish?

The most common are Q4_K_M (14 models), Q8_0 (10 models), F16 (8 models) and Q2_K (3 models).

Keep browsing