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.

Most downloadedNewestLargestLongest contextSmallest
SP
Spark X2.5 4B
Spark · mradermacher
Params4.11B
Context1024K
Best fitQ3_K_M 3.9 GB
GGUF
QW
Qwen3 Coder 30B A3B IT Heretic
Qwen · mradermacher
Params30.5B
Context256K
Q4_K_M18.6 GB
Coding
HE
Hemmingway 1.5
Hemmingway · mradermacher
Params27.3B
Context256K
Q4_K_M16.8 GB
ReasoningTool use
HO
K2-Horizon-3.7B
Horizon · mradermacher
Params5.06B
Context512K
Best fitQ2_K 3.9 GB
GGUF
QW
Qwen3.8 2B Heretic
Qwen · mradermacher
Params1.88B
Context256K
Best fitQ6_K 2.5 GB
ReasoningTool use
QW
knaif
Qwen · blackdeep
Params1.72B
Context40K
Best fitQ6_K 3.1 GB
Tool use
MI
Mistral NeMo 12B Abliterated
Mistral · mradermacher
Params12.2B
Context128K
Q4_K_M7.5 GB
GGUF
LO
Loopl 2b
Loopl · mradermacher
Params1.88B
Context256K
Best fitQ6_K 2.5 GB
Tool useVision
LI
Ling 3.0 flash VL
Ling · bloomer010
Params124B
Context256K
Best fitBF16 3.1 GB
Vision
PH
Phi 4 Mini Reasoning
Phi · bartowski
Params3.84B
Context128K
Best fitQ3_K_XL 4.0 GB
CodingReasoning
QW
Qwen2.5 3B Instruct
Qwen · gaianet
Params3.09B
Context32K
Best fitQ6_K 3.3 GB
GGUF
QW
Qwen3.5 9B Kimi K3 Distilled
Qwen · mradermacher
Params9.2B
Context256K
Q4_K_M5.8 GB
Reasoning
JA
Jais-3-3.7B
Jais · mradermacher
Params5.06B
Context512K
Best fitQ2_K 3.9 GB
Reasoning
LF
LFM3-1.7B
Lfm · lmstudio-community
Params1.73B
Context256K
Best fitQ8_0 4.0 GB
ReasoningTool use
FA
Falcon-H1-1.5B-Deep-Instruct
Falcon · unsloth
Params1.55B
Context128K
Best fitQ8_K_XL 3.0 GB
GGUF
OR
Ornith 1.5 35B A3B DFlash
Ornith · na0x2c6
Params386M
Context256K
Best fitBF16 1.5 GB
GGUF
LL
Llama 3.2 3B Instruct Heretic Ablitered Uncensored
Llama · mradermacher
Params3.21B
Context128K
Best fitQ5_K_M 3.8 GB
GGUF
SM
SmolLM 135M
SmolLM · HuggingFaceTB
Params135M
Context2K
Best fitQ8_0 0.8 GB
Tool use
LO
Loopl 4b
Loopl · mradermacher
Params4.21B
Context256K
Best fitQ3_K_L 4.0 GB
Tool useVision
TO
ToriiGate v0.4 7B
Toriigate · mradermacher
Params7.62B
Context32K
Best fitQ2_K 4.0 GB
Vision
QW
Qwen3 1.7B Abliterated
Qwen · mradermacher
Params1.72B
Context40K
Best fitQ8_0 3.3 GB
GGUF
QW
Qwen2.5 Coder 1.5B Instruct
Qwen · QuantFactory
Params1.54B
Context32K
Best fitQ8_0 2.4 GB
Coding
RW
rwkv7 g1h 1.5b i1
Rwkv · RemySkye
Params1.53B
Context1024K
Best fitQ6_K 2.0 GB
GGUF
HO
K2 Horizon 3.7B Heretic
Horizon · mradermacher
Params5.06B
Context512K
Best fitQ2_K 3.9 GB
GGUF
MI
OpenHermes-2.5-Mistral-7B-x2
Mistral · mradermacher
Params7.24B
Context32K
Best fitIQ2_S 3.9 GB
Tool use
MI
MiniCPM V 2 6
MiniCPM · bartowski
Params7.6B
Context32K
Best fitQ2_K 4.0 GB
CodingVision
QW
Qwythos 9B v2
Qwythos · mradermacher
Params9.2B
Context1024K
Q4_K_M5.8 GB
Reasoning
SK
Skyfall 31B v4.2 Heretic
Skyfall · mradermacher
Params31.4B
Context128K
Q4_K_M19.0 GB
Coding
QW
PopiT Qwen3 4B Medical SFT 1128
Qwen · mradermacher
Params4B
Context40K
Best fitIQ4_XS 4.0 GB
GGUF
PA
Parakeet TDT 0.6B v2 English Hebrew Codeswitching
Parakeet · danielrosehill
Params618M
Context—
Best fitF16 1.9 GB
Coding

Other memory sizes

Keep browsing