GGUF models by MrDevolver

54 GGUF models quantized by MrDevolver on Hugging Face. Sizes run from 3.93B to 48.9B parameters. The longest trained context is 1024K tokens.

Most downloadedNewestLargestLongest contextSmallest
DE
DeepSeek R1 Distill Qwen 14B uncensored Q4 K M
DeepSeek · MrDevolver
Params14.8B
Context128K
Q4_K_M9.0 GB
ReasoningTool use
GP
Gpt Oss 20b Heretic Distilled MiniMax M2.1
gpt-oss · MrDevolver
Params20.9B
Context128K
Smallest14.7 GB
GGUF
GP
Gpt Oss 20b Heretic Distilled MiniMax M2.1
gpt-oss · MrDevolver
Params20.9B
Context128K
Smallest12.1 GB
GGUF
MI
Magistral Geminized
Mistral · MrDevolver
Params23.6B
Context128K
Smallest13.5 GB
ReasoningTool use
NO
NousCoder 14B
Nouscoder · MrDevolver
Params14.8B
Context80K
Q4_K_M9.0 GB
CodingReasoningTool use
PA
PaCoRe 8B
Pacore · MrDevolver
Params8.19B
Context320K
Smallest8.7 GB
GGUF
CE
Celestial Queen 12B Heretic
Celestial · MrDevolver
Params12.2B
Context1000K
Smallest10.1 GB
GGUF
JO
JoyAI LLM Flash
Joyai · MrDevolver
Params48.9B
Context128K
Smallest23.4 GB
GGUF
GP
Gpt Oss Sanguine 20b v1
gpt-oss · MrDevolver
Params20.9B
Context128K
Smallest14.7 GB
CodingReasoningRoleplay
NA
Nanbeige4 3B Thinking 2511
Nanbeige · MrDevolver
Params3.93B
Context64K
Smallest15.7 GB
ReasoningTool use
OP
OpenBuddy R10528Distill 30BA3B Preview2
Openbuddy · MrDevolver
Params30.5B
Context195K
Smallest17.5 GB
ReasoningTool use
AS
Asita 8B
Asita · MrDevolver
Params8.03B
Context128K
Smallest8.5 GB
Tool use
CO
ConsciousCrimininalComputing BlackXorDolphTron FINAL
Consciouscrimininalcomputing · MrDevolver
Params23.6B
Context32K
Smallest13.5 GB
GGUF
DE
DeepSeek R1 0528 Qwen3 8B
DeepSeek · MrDevolver
Params8.19B
Context128K
Smallest8.7 GB
ReasoningTool use
IM
Impish LongPen 12B
Impish · MrDevolver
Params12.2B
Context128K
Q4_K_M7.5 GB
GGUF
IN
InternVL3_5 30B A3B
Internvl · MrDevolver
Params30.5B
Context40K
Smallest17.5 GB
GGUF
QW
Qwen2.5 Recursive Coder v2 14B Instruct
Qwen · MrDevolver
Params14.8B
Context32K
Smallest15.7 GB
CodingTool use
ME
Mergekit Dare_Ties Ftxcrqn
Mergekit · MrDevolver
Params23.6B
Context32K
Smallest13.5 GB
GGUF
ME
mergekit linear ccvlbbk Q4 K M
Mergekit · MrDevolver
Params8.03B
Context128K
Q4_K_M4.9 GB
Tool use
ME
MechaEpstein 8000
Mechaepstein · MrDevolver
Params8.19B
Context40K
Smallest8.7 GB
ReasoningTool use
PR
Prototype X 12b
Prototype · MrDevolver
Params12.2B
Context1000K
Q4_K_M7.5 GB
Roleplay
LL
Llama3.1 SuperDeepFuse CrashCourse12K
Llama · MrDevolver
Params8.03B
Context128K
Smallest6.6 GB
GGUF
TI
TinyR1 32B Preview
Tinyr · MrDevolver
Params32.8B
Context128K
Smallest12.3 GB
ReasoningTool use
KO
Kosmos EVAA immersive mix v45.1 8B Q6 K
Kosmos · MrDevolver
Params8.03B
Context128K
Smallest6.6 GB
GGUF
OP
Open0 2.5
Open · MrDevolver
Params32.8B
Context128K
Smallest12.3 GB
GGUF
QW
InternLM3 8B Instruct Converted Qwen2
Qwen · MrDevolver
Params8.8B
Context32K
Smallest7.2 GB
GGUF
CL
Claumba Average Sft
Claumba · MrDevolver
Params32.6B
Context1024K
Smallest16.7 GB
GGUF
YA
YandexGPT 5 Lite 8B Pretrain
Yandexgpt · MrDevolver
Params8.04B
Context32K
Smallest6.6 GB
GGUF
ME
Mergekit Dare_Ties Fikucxa
Mergekit · MrDevolver
Params23.6B
Context32K
Smallest13.5 GB
GGUF
QW
Openbuddy Qwq 32b v25.1 200k
Qwen · MrDevolver
Params32.8B
Context195K
Smallest12.3 GB
Reasoning

About MrDevolver on Hugging Face

MrDevolver has published 54 GGUF models that ggufmodels.com tracks, across families such as Mergekit, Qwen and gpt-oss. The most downloaded are DeepSeek R1 Distill Qwen 14B uncensored Q4 K M, Gpt Oss 20b Heretic Distilled MiniMax M2.1 and Gpt Oss 20b Heretic Distilled MiniMax M2.1.

The quants MrDevolver publishes most often are Q4_K_M (16), Q4_K_S (12), Q6_K (6) and Q8_0 (5). Their Q4_K_M files range from 4.9 to 10.1 GB. Parameter counts run from 3.93B to 48.9B, and the longest trained context window is 1024K tokens.

Frequently asked questions

What GGUF models has MrDevolver published?

54 models are listed here, mainly from the Mergekit, Qwen and gpt-oss families. The most downloaded is DeepSeek R1 Distill Qwen 14B uncensored Q4 K M.

Which quants does MrDevolver publish?

The most common are Q4_K_M (16 models), Q4_K_S (12 models), Q6_K (6 models) and Q8_0 (5 models).

Keep browsing