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
Ingot 27B DFlash2
Qwen · dendriteholdings
Params1.92B
Context256K
Best fitBF16 4.5 GB
QW
Qwen3.8 27B Med Ablit Heretic
Qwen · mradermacher
Params27B
Context256K
Best fitIQ2_S 12.0 GB
LL
turboderp Cat Llama 3 70B instruct
Llama · RichardErkhov
Params3.3B
Context—
Best fitQ3_K_S 9.7 GB
PH
Phi 3.5 Mini instruct_Uncensored
Phi · bartowski
Params3.82B
Context128K
Best fitF16 11.4 GB
ZE
huggingfaceh4_zephyr-7b-beta
Zephyr · TheBloke
Params7.24B
Context32K
Best fitQ8_0 9.3 GB
GE
Gemma 3 12b It Heretic
Gemma · DreamFast
Params11.8B
Context128K
Best fitQ5_K_S 12.0 GB
QW
Qwen3.8 27B Max Ablit Heretic
Qwen · mradermacher
Params27B
Context256K
Best fitIQ2_S 12.0 GB
ST
stable code 3b
Stable · stabilityai
Params2.8B
Context16K
Smallest2.0 GB
GR
Granite-4.0-H-Micro
Granite · unsloth
Params3.2B
Context1024K
Best fitBF16 9.6 GB
QW
Qwen3 4B Instruct 2507
Qwen · ggml-org
Params4B
Context256K
Best fitQ8_0 6.0 GB
LL
Meta-Llama-3-8B-Instruct-imatrix
Llama · lmstudio-community
Params8B
Context8K
Best fitQ8_0 10.1 GB
MI
Mistral Small 24B Instruct 2501 Abliterated
Mistral · bartowski
Params23.6B
Context32K
Best fitIQ3_XS 11.7 GB
LL
Llama 3.2 4X3B MOE Hell California Uncensored 10B
Llama · DavidAU
Params9.95B
Context128K
Best fitQ8_0 12.0 GB
AB
ABD3ID CodeMentor
Abd · mradermacher
Params8.95B
Context256K
Best fitQ6_K 8.9 GB
DE
DeepSeek R1 Distill Qwen 1.5B
DeepSeek · lmstudio-community
Params1.78B
Context128K
Best fitQ8_0 2.6 GB
DE
deepseek llm 7B chat
DeepSeek · TheBloke
Params6.91B
Context4K
Best fitQ8_0 11.9 GB
LL
Meta Llama 3.1 8B Instruct
Llama · modularai
Params8B
Context128K
Best fitQ6_K 8.2 GB
MY
Mythos 9b Unhinged
Mythos · King3Djbl
Params8.19B
Context40K
Best fitQ8_0 10.4 GB
MI
Magistral-Small-2509
Mistral · unsloth
Params23.6B
Context128K
Best fitIQ3_XXS 11.3 GB
DE
deepseek-ai_deepseek-coder-5.7bmqa-base
DeepSeek · TheBloke
Params5.7B
Context16K
Best fitQ8_0 6.7 GB
CO
Cotype Nano
Cotype · QuantFactory
Params1.54B
Context32K
Best fitQ8_0 2.4 GB
MO
moondream2
Moondream · moondream
Params1.42B
Context2K
Best fitF16 5.0 GB
QW
Swift Qwen3.8 Flash Next
Qwen · bartowski
Params177B
Context256K
Q4_K_M125.6 GB
QW
Qwen3 VL Embedding 8B
Qwen · VesNFF
Params7.6B
Context256K
Best fitQ6_K 7.9 GB
QW
cerebras/Qwen3-Coder-REAP-25B-A3B
Qwen · bartowski
Params25B
Context256K
Best fitIQ3_XS 11.7 GB
PA
PaddleOCR VL manga
Paddleocr · LunarOilRig
Params467M
Context128K
Best fitQ4_K_M 1.0 GB
MI
Mistral Nemo Instruct 2407
Mistral · lmstudio-community
Params12.2B
Context1000K
Best fitQ6_K 11.9 GB
TE
Bonsai-27B
Ternary · Hikari07jp
Params27B
Context256K
Best fitQ2_0 9.8 GB
QW
Qwen3.5 9B Gemini 3.1 Pro Reasoning Distill
Qwen · Jackrong
Params8.95B
Context256K
Best fitQ8_0 11.1 GB
LL
Llama-3-ELYZA-JP-8B
Llama · mmnga
Params8B
Context8K
Best fitQ8_0 10.1 GB