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.
JI
jinen-v1-xsmall-hf
Jinen · togatogah
Params36M
Context1K
Best fitF16 0.7 GB
SE
Sexpedition MS3.2 24B
Sexpedition · mradermacher
Params23.6B
Context128K
Q4_K_M14.3 GB
CO
Confucius4 R2T2
Confucius · netease-youdao
Params1.72B
Context64K
Best fitQ8_0 3.3 GB
TE
Ternary Bonsai 1.7B
Ternary · prism-ml
Params1.72B
Context32K
Smallest0.5 GB
QW
Huihui Qwen3 8B Abliterated v2
Qwen · mradermacher
Params8.19B
Context40K
Best fitIQ1_M 4.0 GB
DA
L3 Dark Planet 8B
Dark · DavidAU
Params8B
Context8K
Best fitQ6_K 2.1 GB
ST
stablelm 2 zephyr 1 6b
Stablelm · stabilityai
Params1.64B
Context4K
Best fitQ8_0 3.9 GB
QW
Qwen2.5 VL 7B Abliterated Caption It
Qwen · mradermacher
Params7.62B
Context125K
Best fitQ2_K 4.0 GB
LL
Hermes 3 Llama 3.2 3b Base Fft Chatml Rerun 8xh100 E4 2024 11 24 4k Steps
Llama · NousResearch
Params3.21B
Context128K
Best fitQ5_K_M 3.8 GB
ON
OneJev 4B
Onejev · mradermacher
Params4.84B
Context256K
Best fitQ3_K_S 3.9 GB
LF
LFM2.5 VL 3B Uncensored
Lfm · SC117
Params2.7B
Context125K
Best fitIQ4_XS 4.0 GB
PH
Phi 4 Mini Reasoning
Phi · lmstudio-community
Params3.84B
Context128K
Best fitQ3_K_L 3.8 GB
QW
Qwen3 14B ARPO DeepSearch
Qwen · mradermacher
Params14.8B
Context40K
Q4_K_M9.0 GB
QW
Qwen3.8-27B-DFlash2-ternary-bonsai2-r3
Qwen · naklitechie
Params1.92B
Context256K
Best fitQ4_K_M 1.8 GB
HO
Holo 3.1 4B
Holo · mradermacher
Params4.84B
Context256K
Best fitQ3_K_S 3.9 GB
TO
tokyotech llm Swallow MS 7b instruct v0.1
Tokyotech · mmnga
Params7.33B
Context4K
Best fitIQ2_S 3.9 GB
QW
qwen2-7b-instruct
Qwen · Qwen
Params7.62B
Context32K
Best fitQ2_K 4.0 GB
QW
Qwen3.5 0.8B GGUF uncensored opus distill
Qwen · amkkk
Params752M
Context256K
Best fitQ4_K_M 1.4 GB
EM
Source Hf Embeddinggemma 2
Embeddinggemma · ngquocvinh
Params271M
Context256K
Best fitQ6_K 2.4 GB
LF
LFM2.5 24B A2B Uncensored Heretic
Lfm · mradermacher
Params24B
Context128K
Q4_K_M14.5 GB
JA
Jais-3-32B
Jais · mradermacher
Params34.8B
Context512K
Q4_K_M21.1 GB
DA
DanTagGen alpha
Dantaggen · KBlueLeaf
Params393M
Context512
Best fitF16 2.1 GB
NA
VisionPsy Nano 460M Flash GGUFs
Nanovlm · qvac
Params409M
Context8K
Q4_K_M0.3 GB
QW
JetBrains Qwen3.8 3.6 27B Blend
Qwen · bartowski
Params27.3B
Context256K
Q4_K_M17.4 GB
SA
Sabi Noema
Sabi · mradermacher
Params27B
Context256K
Q4_K_M16.5 GB
QW
Qwen2 VL 7B Instruct
Qwen · bartowski
Params7.62B
Context32K
Best fitQ2_K 4.0 GB
HE
HERO 35B A3B
Hero · mradermacher
Params34.7B
Context256K
Q4_K_M21.2 GB
DA
DanTagGen delta
Dantaggen · KBlueLeaf
Params393M
Context2K
Best fitF16 2.1 GB
MI
MiniCPM V Apps
MiniCPM · tc-mb
Params912M
Context256K
Best fitF16 2.7 GB
QW
DEV ThinkingCap Qwen3.8 27B
Qwen · ngquocvinh
Params27.3B
Context256K
Q4_K_M16.8 GB