GGUF models that fit in 12 GB
171517 GGUF language models have at least one quant whose estimated memory use is 12 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 4194304K tokens.
QW
Qwen2.5 Coder 1.5B
Qwen · ggml-org
Params1.54B
Context32K
Best fitQ8_0 2.4 GB
IN
Ingot V 27B Q2
Ingot · dendriteholdings
Params26.9B
Context256K
Smallest7.2 GB
OR
RegesCore 1.0 35
Ornith · iBossonline
Params34.7B
Context256K
Best fitIQ1_S 11.7 GB
HE
Hemmingway 1 Heretic MTP V3 Vision
Hemmingway · brainnxdomain
Params27B
Context256K
Best fitBF16 8.6 GB
MI
MegaBeam Mistral 7B 512k
Mistral · bartowski
Params7.24B
Context512K
Best fitQ8_0 9.3 GB
HY
Hy MT2 1.8B 1.25Bit
tencent
Params1.79B
Context256K
Smallest0.5 GB
JE
Jebadiah 9b v2
Jebadiah · bartowski
Params9.2B
Context256K
Best fitQ8_0 11.4 GB
TI
TinyLlama 1.1B Chat v0.6
Tinyllama · TinyLlama
Params1.1B
Context2K
Best fitQ4_0 1.3 GB
MI
ehartford_dolphin-2.1-mistral-7b
Mistral · TheBloke
Params7.24B
Context32K
Best fitQ8_0 9.3 GB
BA
Bathory X 26B A4B 0.1
Bathory · mradermacher
Params26B
Context256K
Q4_K_M17.2 GB
DE
DeepSeek R1 Distill Llama 8B
DeepSeek · mradermacher
Params8B
Context128K
Best fitQ8_0 10.1 GB
GE
Gemma 3 270m Qat Q4_0 Unquantized
Gemma · ggml-org
Params268M
Context32K
Best fitQ4_0 0.9 GB
GO
Goetia 26B A4B v1.7
Goetia · bartowski
Params26B
Context256K
Q4_K_M18.8 GB
QW
Kataguru Sceptic 35B A3B v2.6
Qwen · kataguru
Params35.2B
Context256K
Best fitQ8_0 3.2 GB
GE
gemma4 coding agent
Gemma · S4MPL3BI4S
Params7.5B
Context128K
Best fitQ4_K_M 7.2 GB
LL
Llm Jp 4 8b Instruct
Llm · mmnga-o
Params8.59B
Context64K
Best fitMXFP4_MOE 10.7 GB
QW
Qwen2.5-Omni-3B
Qwen · unsloth
Params3.4B
Context32K
Best fitBF16 7.6 GB
NE
Nemotron 3.5 Bolt E4B GSQ RCO
Nemotron · abenzerps
Params7.52B
Context128K
Best fitQ5_K_M 6.5 GB
CA
CAT Thinking 8B
Cat · mmnga-o
Params8.19B
Context40K
Best fitMXFP4_MOE 10.4 GB
IN
InternLM2
InternLM · legraphista
Params7.74B
Context8K
Best fitQ8_0 9.8 GB
QW
Qwable 9B Claude Fable 5
Qwable · empero-ai
Params8.95B
Context256K
Best fitQ8_0 11.1 GB
NE
NVIDIA Nemotron 3.5 Bolt E4B Pollard
Nemotron · PollardWeights
Params7.52B
Context128K
Best fitQ6_K 6.9 GB
MI
Mistral Small Instruct 2409
Mistral · bartowski
Params22.2B
Context128K
Best fitQ3_K_S 12.0 GB
QW
Qwen2.5 3B Instruct
Qwen · lmstudio-community
Params3.09B
Context32K
Best fitQ8_0 4.1 GB
NA
Nanbeige 4.3 33B Heretic
Nanbeige · Luni
Params33.1B
Context512K
Best fitIQ2_XXS 11.6 GB
QW
Huihui Qwen3 4B Abliterated v2
Qwen · mradermacher
Params4B
Context40K
Best fitQ6_K 5.0 GB
AI
Airoboros L2 7B 2.2
Airoboros · TheBloke
Params6.74B
Context4K
Best fitQ8_0 12.0 GB
QW
Qwen3 4B Thinking 2507
Qwen · bartowski
Params4B
Context256K
Best fitBF16 9.8 GB
DE
DeepSeek V4.1 Flash 1017 ELVEZ Q2
DeepSeek · elvezjp
Params485M
Context—
Smallest1.0 GB
HU
Hunyuan MT 7B
Hunyuan · imikeliu
Params7.5B
Context256K
Best fitQ4_K_M 6.2 GB