Q3_K_S GGUF models

3-bit k-quant, small. Small files with noticeable quality loss. 102724 models have a Q3_K_S file. Sizes run from 1M to 2780B parameters. The longest trained context is 4194304K tokens.

Most downloadedNewestLargestLongest contextSmallest
MA
MN 12B Mag Mell R1 Uncensored Scale1.2 i1
Mag · mradermacher
Params12.2B
Context1000K
Q4_K_M7.5 GB
Reasoning
LT
LTX2.3 10Eros 1.4
Ltx · vantagewithai
Params21B
Context—
Q4_K_M14.3 GB
GGUF
GE
Gemma 4 E4B It Uncensored
Gemma · mradermacher
Params7.52B
Context128K
Q4_K_M5.3 GB
ReasoningTool use
DA
Darwin 36B Opus ABLITERATED HERETIC
Darwin · mradermacher
Params34.7B
Context256K
Q4_K_M21.2 GB
GGUF
MI
OpenHermes 2.5 Mistral 7B Orca DPO
Mistral · mradermacher
Params7.24B
Context32K
Q4_K_M4.4 GB
Tool use
GL
Glm Oss 8b
GLM · mradermacher
Params8B
Context128K
Q4_K_M4.9 GB
GGUF
QW
Josiefied Qwen3 8B Abliterated v1
Qwen · mradermacher
Params8.19B
Context40K
Q4_K_M5.0 GB
GGUF
RU
RusMind Malysh 4.7
Rusmind · mradermacher
Params752M
Context256K
Q4_K_M0.5 GB
ReasoningTool use
QW
Qwen3.5 9b Triple Loop Fase2
Qwen · mradermacher
Params8.95B
Context256K
Q4_K_M5.6 GB
ReasoningTool use
MI
Dolphin Mistral 24B Venice Edition
Mistral · mradermacher
Params23.6B
Context32K
Q4_K_M14.3 GB
GGUF
YI
dolphin-2.2-yi-34b-200k
Yi · mradermacher
Params34.4B
Context195K
Q4_K_M20.7 GB
GGUF
AS
Aspire 8B model stock i1
Aspire · mradermacher
Params8B
Context8K
Q4_K_M4.9 GB
GGUF
SW
Swallow 70b Instruct v0.1
Swallow · RichardErkhov
Params69B
Context4K
Q4_K_M41.5 GB
GGUF
GR
Granite 3.1 3b A800M Instruct
Granite · bartowski
Params3.3B
Context128K
Q4_K_M2.0 GB
GGUF
QW
Heretic Qwen3.5 4b
Qwen · mradermacher
Params4.21B
Context256K
Q4_K_M2.7 GB
ReasoningTool use
TU
Tulu2 7b_writing Prompts
Tulu · RichardErkhov
Params6.74B
Context4K
Q4_K_M4.1 GB
GGUF
MI
danielbrdz_barcenas-mistral-7b
Mistral · TheBloke
Params7.24B
Context32K
Q4_K_M4.4 GB
GGUF
MI
rana shahroz mistral 7b openorca20k
Mistral · RichardErkhov
Params7.24B
Context32K
Q4_K_M4.4 GB
GGUF
GE
gemma-3-27b-it-codeforces-SFT
Gemma · bartowski
Params27B
Context128K
Q4_K_M16.5 GB
Reasoning
QW
Qwen2.5 7B
Qwen · mradermacher
Params7.62B
Context128K
Q4_K_M4.7 GB
GGUF
QW
Qwen2.5 3B Instruct Abliterated
Qwen · mradermacher
Params3.09B
Context32K
Q4_K_M1.9 GB
GGUF
GR
Granite 8b Code Base 4k
Granite · RichardErkhov
Params8.05B
Context4K
Q4_K_M4.9 GB
Coding
WI
WizardCoder 33B V1.1
Wizardcoder · mradermacher
Params33.3B
Context16K
Q4_K_M19.9 GB
Coding
ZE
zed industries zeta 2
Zeta · bartowski
Params8.25B
Context32K
Q4_K_M5.2 GB
GGUF
QW
Qwen3 4B Base
Qwen · mradermacher
Params4B
Context32K
Q4_K_M2.5 GB
GGUF
LL
CodeLlama 70b Instruct hf
Llama · second-state
Params69B
Context4K
Q4_K_M41.4 GB
Coding
MI
Dolphin3.0 Mistral 24B
Mistral · mradermacher
Params23.6B
Context32K
Q4_K_M14.3 GB
CodingTool use
OL
open r1 OlympicCoder 7B
Olympiccoder · bartowski
Params7.62B
Context32K
Q4_K_M4.7 GB
CodingReasoning
VI
Vinci MLE 8B 1.0
Vinci · mradermacher
Params8.79B
Context128K
Q4_K_M5.3 GB
Tool use
CU
CursorCore QW2.5 1.5B i1
Cursorcore · mradermacher
Params1.54B
Context32K
Q4_K_M1.0 GB
Coding

Other quants

Keep browsing