GGUF models that fit in 4 GB
79024 GGUF language models have at least one quant whose estimated memory use is 4 GB or less at 8K context. Each card shows the largest quant that fits. Sizes run from 1M to 20260731B parameters. The longest trained context is 1280000K tokens.
NE
NeoHorse 1 9 Uncensored
Neohorse · mradermacher
Params8.95B
Context256K
Q4_K_M5.6 GB
DA
Dark Scarlett 27B v3.0
Dark · ReadyArt
Params27.3B
Context256K
Q4_K_M23.5 GB
EK
Ekbasis 27B
Ekbasis · mradermacher
Params27B
Context256K
Q4_K_M16.5 GB
GE
gemmuguinho 636m
Gemmuguinho · Papuguinho
Params636M
Context32K
Best fitF16 1.9 GB
KA
KansenSakura Eclipse RP 12b
Kansensakura · mradermacher
Params12.2B
Context1000K
Q4_K_M7.5 GB
LL
Llama 3.2 3B Instruct
Llama · mradermacher
Params3.61B
Context128K
Best fitQ5_K_S 4.0 GB
GE
gemma 4 E2B it UD japanese imatrix
Gemma · dahara1
Params4.65B
Context128K
Best fitQ3_K_XL 4.0 GB
MI
MiniCPM V 4.7 1B Heretic
MiniCPM · mradermacher
Params1.01B
Context256K
Best fitQ6_K 1.8 GB
QW
Qwen3-Vl-8B-Thinking-1M
Qwen · unsloth
Params8.19B
Context977K
Best fitIQ1_S 4.0 GB
HY
HyperNova 60B
Hypernova · mradermacher
Params58.7B
Context128K
Q4_K_M40.7 GB
QW
WhereOPD Qwen3.5 4B
Qwen · mradermacher
Params4.84B
Context256K
Best fitIQ3_S 4.0 GB
LL
Llama-3-Instruct-Neurona-8b-v2
Llama · mradermacher
Params8B
Context8K
Best fitIQ2_XXS 4.0 GB
LF
LFM2.5 24B A2B Abliterated
Lfm · mradermacher
Params24B
Context128K
Q4_K_M14.5 GB
ST
L3-8B-Stheno-v3.2
Stheno · DavidAU
Params8B
Context8K
Best fitIQ1_M 3.7 GB
DE
DeepNarra-Paham-Experimental
Deepnarra · DeepNarra
Params952M
Context2K
Smallest0.3 GB
DE
Qwen3 30B A3B Deepseek Distill Instruct 2507
DeepSeek · mradermacher
Params30.5B
Context256K
Q4_K_M18.6 GB
MI
MiniCPM5-2B-Claude-Fable5-1-Thinking
MiniCPM · GnLOLot
Params2.52B
Context128K
Best fitQ8_0 3.5 GB
DE
Qwen2.5 MOE 2X1.5B DeepSeek Uncensored Censored 4B
DeepSeek · DavidAU
Params4.1B
Context128K
Best fitQ5_K_M 3.7 GB
LL
Meta Llama 3 8B Instruct GGUF Imatrix smashed
Llama · PrunaAI
Params8B
Context8K
Best fitIQ2_XXS 4.0 GB
QW
dnotitia_Smoothie Qwen3 4B
Qwen · Alcoft
Params4B
Context40K
Best fitQ4_K_S 4.0 GB
BO
Borealis Open 12b
Borealis · mradermacher
Params11.8B
Context128K
Q4_K_M7.3 GB
LF
LFM2.5 8B A1B Hermes Agentic Coder Abliterated v3
Lfm · mradermacher
Params8.47B
Context125K
Best fitIQ1_S 3.9 GB
QW
DPO Trained Qwen2.5 Python Coder 7B
Qwen · mradermacher
Params7.62B
Context32K
Best fitQ2_K 4.0 GB
CH
Chaotic Order 24B V5
Chaotic · mradermacher
Params23.6B
Context128K
Q4_K_M14.3 GB
QW
Qwen3.5 2B Uncensored Zero Refusal SOMA
Qwen · mradermacher
Params1.88B
Context256K
Best fitQ6_K 2.5 GB
QW
Qwen3 Coder Next Uncensored Heretic
Qwen · mradermacher
Params79.7B
Context256K
Q4_K_M48.5 GB
YU
Ruach Studio Models v2
Yue · goldhub
Params632M
Context32K
Best fitQ6_K 3.9 GB
CO
Yi Coder 1.5B
Coder · lmstudio-community
Params1.48B
Context128K
Best fitQ8_0 3.7 GB
NE
Nemotron 3.5 Embed 8B BF16
Nemotron · W4D
Params7.95B
Context256K
Best fitIQ2_XXS 4.0 GB
RE
REBEL-OpenChat-3.5
Rebel · mradermacher
Params7.24B
Context8K
Best fitIQ2_S 3.9 GB