GGUF models by tinyopsec

111 GGUF models quantized by tinyopsec on Hugging Face. Sizes run from 4M to 13B parameters. The longest trained context is 256K tokens.

Most downloadedNewestLargestLongest contextSmallest
QW
Qwen3.5 4B Uncensored Aggressive
Qwen · tinyopsec
Params4.33B
Context256K
Q4_K_M2.8 GB
GGUF
LL
Llama 3.2 3B Instruct heretic ablitered uncensored
Llama · tinyopsec
Params3.2B
Context128K
Q4_K_M2.0 GB
GGUF
QW
Qwen3 1.7B Coder Distilled SFT
Qwen · tinyopsec
Params2.03B
Context40K
Q4_K_M1.3 GB
CodingReasoningTool use
QW
Qwen3.8 2B Function Calling xLAM
Qwen · tinyopsec
Params1.94B
Context256K
Q4_K_M1.3 GB
ReasoningTool use
MI
pjev 2b
MiniCPM · tinyopsec
Params2.52B
Context128K
Q4_K_M1.6 GB
ReasoningTool use
LL
Vikhr-Llama-3.2-1B-instruct
Llama · tinyopsec
Params1.24B
Context128K
Q4_K_M0.8 GB
Tool use
MI
MiniCPM5 2B
MiniCPM · tinyopsec
Params2.52B
Context128K
Q4_K_M1.6 GB
ReasoningTool use
LL
llama3.2 1b CoT
Llama · tinyopsec
Params1.24B
Context128K
Q4_K_M0.8 GB
ReasoningTool use
MI
MiniCPM5 2B Abliterated Uncensored Safetensors
MiniCPM · tinyopsec
Params2.52B
Context128K
Q4_K_M1.6 GB
ReasoningTool use
LL
Dolphin 3.0 Llama 3.2 1b
Llama · tinyopsec
Params1.24B
Context128K
Q4_K_M0.8 GB
Tool use
DU
DualMind TKD Agentic 1.7B
Dualmind · tinyopsec
Params1.72B
Context40K
Q4_K_M1.1 GB
ReasoningTool use
NA
Nanbeige4.2 3B
Nanbeige · tinyopsec
Params4.17B
Context256K
Q4_K_M2.6 GB
ReasoningTool use
SK
Skywork OR1 Math 7B
Skywork · tinyopsec
Params7.62B
Context128K
Q4_K_M4.7 GB
ReasoningTool use
PY
Pythagoras Prover 4B
Pythagoras · tinyopsec
Params4.41B
Context40K
Q4_K_M2.7 GB
ReasoningTool use
QW
Qwen3 1.7B Distilled 30B A3B
Qwen · tinyopsec
Params2.03B
Context40K
Q4_K_M1.3 GB
ReasoningTool use
LL
Llama 3.2 1B
Llama · tinyopsec
Params1.07B
Context128K
Q4_K_M1.0 GB
GGUF
SV
Jeesup svd safety l3 swift remove30 swapgapiter evfront b010
Svd · tinyopsec
Params8.03B
Context8K
Q4_K_M4.9 GB
GGUF
JE
Jeesup safety l3 swift remove20 swapgapiter b010
Jeesup · tinyopsec
Params8.03B
Context8K
Q4_K_M4.9 GB
GGUF
SV
Jeesup svd safety l3 swift jbbcal2 remove40
Svd · tinyopsec
Params8.03B
Context8K
Q4_K_M4.9 GB
GGUF
LF
LFM2.5 2.6B TauBanking SFT 01
Lfm · tinyopsec
Params2.7B
Context128K
Q4_K_M1.7 GB
ReasoningTool use
GE
GenPRM 7B
Genprm · tinyopsec
Params7.62B
Context128K
Q4_K_M4.7 GB
ReasoningTool use
QW
Nemotron Research GooseReason 4B Instruct
Qwen · tinyopsec
Params4.41B
Context256K
Q4_K_M2.7 GB
CodingReasoningTool use
SV
Jeesup svd safety l3 swift jbbcal2 remove50
Svd · tinyopsec
Params8.03B
Context8K
Q4_K_M4.9 GB
GGUF
MI
minicpm5 2b distilled reasoning
MiniCPM · tinyopsec
Params2.52B
Context128K
Q4_K_M1.6 GB
ReasoningTool use
QW
novel humanize 4b
Qwen · tinyopsec
Params4.02B
Context40K
Q4_K_M2.5 GB
ReasoningTool use
QW
bible assistant qwen35 4b
Qwen · tinyopsec
Params4.21B
Context256K
Q4_K_M2.7 GB
ReasoningTool use
LF
Pallete 2.6B RP Reasoning 2609 v1b
Lfm · tinyopsec
Params2.7B
Context128K
Q4_K_M1.7 GB
ReasoningRoleplayTool use
LL
SimNPO WMDP llama3 8b
Llama · tinyopsec
Params8.03B
Context8K
Q4_K_M4.9 GB
GGUF
QW
Arch Router 1.5B
Qwen · tinyopsec
Params1.54B
Context32K
Q4_K_M1.0 GB
Tool use
QW
switchboard router 4b
Qwen · tinyopsec
Params4.02B
Context256K
Q4_K_M2.5 GB
Tool use

About tinyopsec on Hugging Face

tinyopsec has published 111 GGUF models that ggufmodels.com tracks, across families such as Qwen, Llama and MiniCPM. The most downloaded are Qwen3.5 4B Uncensored Aggressive, Llama 3.2 3B Instruct heretic ablitered uncensored and Qwen3 1.7B Coder Distilled SFT.

The quants tinyopsec publishes most often are F16 (116), Q8_0 (114), Q5_K_M (114) and Q4_K_M (114). Their Q4_K_M files range from 0.0 to 7.9 GB. Parameter counts run from 4M to 13B, and the longest trained context window is 256K tokens.

Frequently asked questions

What GGUF models has tinyopsec published?

111 models are listed here, mainly from the Qwen, Llama and MiniCPM families. The most downloaded is Qwen3.5 4B Uncensored Aggressive.

Which quants does tinyopsec publish?

The most common are F16 (116 models), Q8_0 (114 models), Q5_K_M (114 models) and Q4_K_M (114 models).

Keep browsing