AI models that fit on a 8 GB GPU

Open models that fit in 8 GB of VRAM: 540 as published, 704 at FP8 and 1048 at INT4. Each model is listed on the smallest tier it fits at that precision.

GPUs with 8 GB

Cards whose datasheet VRAM puts them in this tier, up to the next tier at 12 GB. Prices are the lowest live data-center on-demand rate per GPU.

GPUVRAMFrom per GPU hour
RTX 3060 Ti8 GBNo live offer
RTX 30708 GBNo live offer
RTX 3070 Ti8 GBNo live offer
RTX 40608 GBNo live offer
RTX 50608 GBNo live offer
RTX 308010 GBNo live offer
GTX 1080 Ti11 GBNo live offer
RTX 2080 Ti11 GBNo live offer

Fits as published

Models whose published weights, plus the flat overhead, fit this much memory with no quantization.

Text models

The 40 most downloaded of 244.

ModelParametersPublished asVRAM needed
Qwen3-0.6B752MBF161.7 GB
gpt2137MF320.6 GB
Qwen2.5-1.5B-Instruct1.5BBF163.5 GB
Qwen2.5-3B-Instruct3.1BBF166.9 GB
Llama-3.2-1B-Instruct1.2BBF162.8 GB
Qwen2.5-0.5B-Instruct494MBF161.1 GB
Qwen3-1.7B2.0BBF164.5 GB
pythia-160m213MF160.5 GB
gemma-3-1b-it1000MBF162.2 GB
Kimi-K3-DSpark2.2BBF165.0 GB
SmolLM2-135M135MBF160.3 GB
Bonsai-27B-mlx-1bit1.7BF163.9 GB
Qwen3-1.7B-Base1.7BBF163.8 GB
Qwen2.5-0.5B494MBF161.1 GB
phi-22.8BF166.2 GB
Llama-3.2-3B-Instruct3.2BBF167.2 GB
SmolLM2-135M-Instruct135MBF160.3 GB
gemma-3-270m268MBF160.6 GB
OpenELM-1_1B-Instruct1.1BBF162.4 GB
Llama-3.2-1B1.2BBF162.8 GB
Qwen3-4B-Instruct-2507-FP84.4BF8_E4M34.9 GB
Qwen2.5-1.5B1.5BBF163.5 GB
gpt2-large812MF323.6 GB
OLMo-2-0425-1B1.5BF326.6 GB
Qwen3-0.6B-Base596MBF161.3 GB
bloomz-560m559MF161.2 GB
Qwen2-1.5B-Instruct1.5BBF163.5 GB
Qwen2-0.5B494MBF161.1 GB
Qwen2.5-Math-1.5B1.5BBF163.5 GB
MiniCPM5-1B1.1BBF162.4 GB
Qwen2.5-Coder-1.5B-Instruct1.5BBF163.5 GB
gemma-2-2b-it2.6BBF165.8 GB
SmolLM3-3B-Base3.1BBF166.9 GB
pythia-160m-deduped213MF160.5 GB
SmolLM3-3B3.1BBF166.9 GB
japanese-gpt-neox-small204MF320.9 GB
macbert4csc-base-chinese102MF320.5 GB
gpt-neo-125m150MF320.7 GB
DeepSeek-R1-Distill-Qwen-1.5B1.8BBF164.0 GB
bloom-560m559MF161.2 GB

Vision-language models

The 40 most downloaded of 60.

ModelParametersPublished asVRAM needed
Unlimited-OCR3.3BBF167.5 GB
Qwen3-VL-2B-Instruct2.1BBF164.8 GB
Qwen3.5-2B2.3BBF165.1 GB
Florence-2-base232MF160.5 GB
DeepSeek-OCR3.3BBF167.5 GB
Qwen3.5-0.8B873MBF162.0 GB
GLM-OCR1.3BBF163.0 GB
Qwen2-VL-2B-Instruct2.2BBF164.9 GB
moondream21.9BBF164.3 GB
SmolVLM2-500M-Video-Instruct507MF322.3 GB
Qwen3.5-2B-Base2.3BBF165.1 GB
surya-ocr-2686MBF161.5 GB
DeepSeek-OCR-23.4BBF167.6 GB
Cosmos-Reason2-2B2.4BBF165.5 GB
Rax-4.52.3BBF165.1 GB
GOT-OCR2_0716MBF161.6 GB
HunyuanOCR1.1BBF162.5 GB
Florence-2-large777MF161.7 GB
llava-onevision-qwen2-0.5b-ov-hf894MF162.0 GB
InternVL2-2B2.2BBF164.9 GB
InternVL2-1B938MBF162.1 GB
SmolVLM-256M-Instruct256MBF160.6 GB
MiniCPM-V-4.61.3BBF162.9 GB
dots.mocr3.0BBF166.8 GB
Florence-2-large777MF161.7 GB
Unlimited-OCR-23.3BBF167.5 GB
LightOnOCR-2-1B1.0BBF162.2 GB
dots.ocr3.0BBF166.8 GB
MinerU2.5-Pro-2604-1.2B1.2BBF162.6 GB
deepseek-vl2-tiny3.4BBF167.5 GB
LFM2.5-VL-1.6B1.6BBF163.6 GB
Qwen3.5-0.8B-Base873MBF162.0 GB
Florence-2-VQAJP2271MBF160.6 GB
Florence-2-base-ft232MF160.5 GB
Qwen3-VL-4B-Instruct-FP84.8BF8_E4M35.4 GB
granite-docling-258M258MBF160.6 GB
InternVL3-1B938MBF162.1 GB
GOT-OCR-2.0-hf561MBF161.3 GB
Qari-OCR-v0.3-VL-2B-Instruct2.2BF164.9 GB
InternVL3-1B-hf938MBF162.1 GB

Image generation models

The 40 most downloaded of 47.

Video generation models

10 models.

ModelParametersPublished asVRAM needed
Wan2.1-T2V-1.3B-Diffusers1.4BF326.3 GB
stable-video-diffusion-img2vid-xt1.5BF326.8 GB
TurboWan2.1-T2V-1.3B-Diffusers1.4BBF163.2 GB
stable-video-diffusion-img2vid1.5BF326.8 GB
i2vgen-xl1.4BF326.3 GB
Wan2.1-T2V-1.3B1.4BF326.3 GB
CogVideoX-2b1.7BF163.8 GB
text-to-video-ms-1.7b1.4BF326.3 GB
LTX-Video-0.9.51.9BBF164.3 GB
stable-virtual-camera1.3BF325.7 GB

Speech models

The 40 most downloaded of 124.

Embedding models

10 models.

ModelParametersPublished asVRAM needed
Qwen3-Embedding-0.6B596MBF161.3 GB
multilingual-e5-large-instruct560MF161.3 GB
Qwen3-VL-Embedding-2B2.1BBF164.8 GB
gte-Qwen2-1.5B-instruct1.8BF327.9 GB
embeddinggemma-2744MBF161.7 GB
voyage-4-nano346MBF160.8 GB
jina-code-embeddings-0.5b494MBF161.1 GB
stella_en_1.5B_v51.5BF326.9 GB
jina-code-embeddings-1.5b1.5BBF163.5 GB
WeMM-Embedding-2B2.7BBF166.1 GB

Other models

The 40 most downloaded of 45.

ModelParametersPublished asVRAM needed
bert-base-uncased110MF320.5 GB
sam3860MF323.8 GB
Qwen3-VL-Reranker-2B2.1BBF164.8 GB
Qwen3-Reranker-0.6B596MBF161.3 GB
turn-detector135MF320.6 GB
dinov3-vitl16-pretrain-lvd1689m303MF321.4 GB
gemma-4-31B-it-assistant470MBF161.0 GB
blip-image-captioning-large470MF322.1 GB
MOSS-Transcribe-Diarize909MBF162.0 GB
trocr-base-printed333MF321.5 GB
kosmos-2-patch14-2241.7BF327.4 GB
gemma-4-26B-A4B-it-assistant420MBF160.9 GB
trocr-base-handwritten333MF321.5 GB
LightOnOCR-1B-10251.2BBF162.6 GB
mxbai-rerank-large-v21.5BF163.5 GB
GLiNER2.5-Decide486MF322.2 GB
dinov3-vith16plus-pretrain-lvd1689m841MF323.8 GB
gemma-4-12B-it-qat-q4_0-unquantized-assistant423MBF160.9 GB
sundial-base-128m128MF320.6 GB
mxbai-rerank-base-v2494MF161.1 GB
gemma-4-12B-it-assistant423MBF160.9 GB
Janus-Pro-1B2.1BBF164.6 GB
Timestral-209M209MF320.9 GB
laya421MF160.9 GB
stable-fast-3d1.0BF324.5 GB
fg-clip-base150MF320.7 GB
LFM3-Omni-2B2.2BBF164.8 GB
texture-fix-vae-for-qwen-image-2.1338MF321.5 GB
gliner2.5-multi-v1287MF321.3 GB
pixai-tagger-v1.0486MF322.2 GB
GLiNER2.5-multi-Decide287MF321.3 GB
LFM3-Audio-2B2.0BBF164.5 GB
muscriptor-large1.4BF326.1 GB
OXR-1.01.5BBF163.4 GB
Julia-1144MF320.6 GB
Alpamayo2-Edge2.7BBF165.9 GB
laya-multilingual322MF160.7 GB
LeVJEPA-VideoMix-Large303MF321.4 GB
ARC-1.0615MBF161.4 GB
MiniMaxImage_hdr_diffusers_vae584MF322.6 GB

Fits at FP8

Models that fit only after quantizing the weights to 8 bits (1 byte per parameter). Needs an FP8 checkpoint or an engine that quantizes on load, and a GPU with FP8 support.

Text models

The 40 most downloaded of 310.

ModelParametersPublished asVRAM needed
Qwen3-0.6B752MBF160.8 GB
gpt2137MF320.2 GB
Qwen2.5-1.5B-Instruct1.5BBF161.7 GB
Qwen2.5-3B-Instruct3.1BBF163.4 GB
Llama-3.2-1B-Instruct1.2BBF161.4 GB
Qwen3-4B4.0BBF164.5 GB
Qwen2.5-0.5B-Instruct494MBF160.6 GB
Qwen3-1.7B2.0BBF162.3 GB
pythia-160m213MF160.2 GB
Qwen3-4B-Instruct-25074.0BBF164.5 GB
gemma-3-1b-it1000MBF161.1 GB
Kimi-K3-DSpark2.2BBF162.5 GB
SmolLM2-135M135MBF160.2 GB
Bonsai-27B-mlx-1bit1.7BF161.9 GB
Qwen3-1.7B-Base1.7BBF161.9 GB
Qwen3-4B-Base4.0BBF164.5 GB
Qwen2.5-0.5B494MBF160.6 GB
phi-22.8BF163.1 GB
Llama-3.2-3B-Instruct3.2BBF163.6 GB
SmolLM2-135M-Instruct135MBF160.2 GB
gemma-3-270m268MBF160.3 GB
OpenELM-1_1B-Instruct1.1BBF161.2 GB
Llama-3.2-1B1.2BBF161.4 GB
PowerMoE-3b3.4BF323.8 GB
Qwen2.5-1.5B1.5BBF161.7 GB
gpt2-large812MF320.9 GB
OLMo-2-0425-1B1.5BF321.7 GB
Qwen3-0.6B-Base596MBF160.7 GB
bloomz-560m559MF160.6 GB
Qwen2-1.5B-Instruct1.5BBF161.7 GB
Qwen2-0.5B494MBF160.6 GB
Qwen2.5-Math-1.5B1.5BBF161.7 GB
deepseek-coder-7b-instruct-v1.56.9BBF167.7 GB
Llama-2-7b-hf6.7BF167.5 GB
MiniCPM5-1B1.1BBF161.2 GB
Qwen2.5-Coder-1.5B-Instruct1.5BBF161.7 GB
gemma-2-2b-it2.6BBF162.9 GB
SmolLM3-3B-Base3.1BBF163.4 GB
pythia-160m-deduped213MF160.2 GB
Phi-3-mini-4k-instruct3.8BBF164.3 GB

Vision-language models

The 40 most downloaded of 93.

ModelParametersPublished asVRAM needed
Qwen3.5-4B4.7BBF165.2 GB
Qwen3-VL-4B-Instruct4.4BBF165.0 GB
Qwen2.5-VL-3B-Instruct3.8BBF164.2 GB
Unlimited-OCR3.3BBF163.7 GB
chandra-ocr-25.3BBF165.9 GB
Qwen3-VL-2B-Instruct2.1BBF162.4 GB
Qwen3.5-2B2.3BBF162.5 GB
Florence-2-base232MF160.3 GB
DeepSeek-OCR3.3BBF163.7 GB
Qwen3.5-0.8B873MBF161.0 GB
llava-1.5-7b-hf7.1BF167.9 GB
GLM-OCR1.3BBF161.5 GB
Qwen2-VL-2B-Instruct2.2BBF162.5 GB
moondream21.9BBF162.2 GB
gemma-3-4b-it4.3BBF164.8 GB
SmolVLM2-500M-Video-Instruct507MF320.6 GB
Qwen3.5-2B-Base2.3BBF162.5 GB
surya-ocr-2686MBF160.8 GB
DeepSeek-OCR-23.4BBF163.8 GB
medgemma-4b-it4.3BBF164.8 GB
Cosmos-Reason2-2B2.4BBF162.7 GB
Rax-4.52.3BBF162.5 GB
GOT-OCR2_0716MBF160.8 GB
Phi-3.5-vision-instruct4.1BBF164.6 GB
HunyuanOCR1.1BBF161.3 GB
Florence-2-large777MF160.9 GB
llava-onevision-qwen2-0.5b-ov-hf894MF161.0 GB
InternVL2-2B2.2BBF162.5 GB
InternVL2-1B938MBF161.0 GB
SmolVLM-256M-Instruct256MBF160.3 GB
MiniCPM-V-4.61.3BBF161.5 GB
dots.mocr3.0BBF163.4 GB
Florence-2-large777MF160.9 GB
blip2-opt-2.7b3.7BF324.2 GB
Qwen3.5-4B-Base4.7BBF165.2 GB
Unlimited-OCR-23.3BBF163.7 GB
LightOnOCR-2-1B1.0BBF161.1 GB
typhoon-ocr-3b3.8BBF164.2 GB
Nanonets-OCR-s3.8BBF164.2 GB
vllm-translategemma-4b-it5.0BBF165.6 GB

Image generation models

The 40 most downloaded of 71.

ModelParametersPublished asVRAM needed
stable-diffusion-xl-base-1.02.6BF322.9 GB
stable-diffusion-v1-5860MF321.0 GB
dreamshaper-7860MF321.0 GB
sdxl-turbo2.6BF322.9 GB
Z-Image-Turbo6.2BF326.9 GB
stable-diffusion-v1-4860MF321.0 GB
RealVisXL_V5.02.6BF322.9 GB
sd-turbo866MF321.0 GB
playground-v2.5-1024px-aesthetic2.6BF322.9 GB
Qwen-Image-2.1-viggle-turbo7.1BBF168.0 GB
animagine-xl-4.02.6BF162.9 GB
animagine-xl-3.12.6BF162.9 GB
nova-furry-xl-il-v120-sdxl2.6BF162.9 GB
one-obsession-17-red-sdxl2.6BF162.9 GB
dreamshaper-8860MF321.0 GB
stable-diffusion-3.5-medium2.5BBF162.8 GB
amanatsu-illustrious-v11-sdxl2.6BF162.9 GB
controlnet-union-sdxl-1.01.3BF161.4 GB
LCM_Dreamshaper_v7860MF321.0 GB
diving-illustrious-real-asian-v50-sdxl2.6BF162.9 GB
controlnet-openpose-sdxl-1.01.3BF161.4 GB
noobai-XL-1.12.6BF162.9 GB
stable-diffusion-xl-1.0-inpainting-0.12.6BF322.9 GB
Qwen-Image-2.17.1BBF168.0 GB
janku-v5-nsfw-trained-noobai-rou-wei-illustrious-xl-v50-sdxl2.6BF162.9 GB
Illustrious-xl-early-release-v02.6BF162.9 GB
noobai-XL-Vpred-1.02.6BF162.9 GB
obsession-illustriousxl-v10-sdxl2.6BF162.9 GB
animagine-xl-3.02.6BF162.9 GB
hassaku-xl-illustrious-v31-sdxl2.6BF162.9 GB
controlnet-scribble-sdxl-1.01.3BF161.4 GB
Z-Image6.2BBF166.9 GB
prefect-illustrious-xl-v3-sdxl2.6BF162.9 GB
Qwen-Image-2.1-Turbo7.1BBF168.0 GB
Photon_v1860MF321.0 GB
stable-diffusion-3-medium-diffusers2.1BF162.3 GB
stable-diffusion-2-1-base866MF321.0 GB
minecraft-skin-generator-sdxl2.6BF322.9 GB
autismmix-sdxl-autismmix-pony-sdxl2.6BF162.9 GB
true-pencil-xl-v100-sdxl2.6BF162.9 GB

Video generation models

17 models.

ModelParametersPublished asVRAM needed
LTX-Video1.9BF322.1 GB
Wan2.1-T2V-1.3B-Diffusers1.4BF321.6 GB
stable-video-diffusion-img2vid-xt1.5BF321.7 GB
Wan2.2-TI2V-5B-Diffusers5.0BF325.6 GB
FastWan2.2-TI2V-5B-FullAttn-Diffusers5.0BBF165.6 GB
TurboWan2.1-T2V-1.3B-Diffusers1.4BBF161.6 GB
stable-video-diffusion-img2vid1.5BF321.7 GB
i2vgen-xl1.4BF321.6 GB
Wan2.1-T2V-1.3B1.4BF321.6 GB
CogVideoX-2b1.7BF161.9 GB
text-to-video-ms-1.7b1.4BF321.6 GB
LTX-Video-0.9.51.9BBF162.1 GB
CogVideoX-5b5.6BBF166.2 GB
LongLive-2.0-5B-Diffusers5.0BBF165.6 GB
CogVideoX-5b-I2V5.6BBF166.3 GB
Wan2.1-VACE-1.3B-diffusers2.2BF322.4 GB
stable-virtual-camera1.3BF321.4 GB

Speech models

The 40 most downloaded of 139.

Embedding models

11 models.

ModelParametersPublished asVRAM needed
Qwen3-Embedding-0.6B596MBF160.7 GB
Qwen3-Embedding-4B4.0BBF164.5 GB
multilingual-e5-large-instruct560MF160.6 GB
Qwen3-VL-Embedding-2B2.1BBF162.4 GB
gte-Qwen2-1.5B-instruct1.8BF322.0 GB
embeddinggemma-2744MBF160.8 GB
voyage-4-nano346MBF160.4 GB
jina-code-embeddings-0.5b494MBF160.6 GB
stella_en_1.5B_v51.5BF321.7 GB
jina-code-embeddings-1.5b1.5BBF161.7 GB
WeMM-Embedding-2B2.7BBF163.0 GB

Other models

The 40 most downloaded of 63.

ModelParametersPublished asVRAM needed
bert-base-uncased110MF320.1 GB
gemma-4-E2B-it5.1BBF165.7 GB
Qwen3-Reranker-4B4.0BBF164.5 GB
sam3860MF321.0 GB
Qwen3-VL-Reranker-2B2.1BBF162.4 GB
Qwen3-Reranker-0.6B596MBF160.7 GB
turn-detector135MF320.2 GB
dinov3-vitl16-pretrain-lvd1689m303MF320.3 GB
FLUX.2-klein-4B3.9BBF164.3 GB
gemma-4-31B-it-assistant470MBF160.5 GB
blip-image-captioning-large470MF320.5 GB
gemma-4-E2B-it-qat-w4a16-ct5.6BBF166.2 GB
bge-reranker-v2-gemma2.5BF322.8 GB
Qwen3.5-4B-3D-Coder4.5BF165.1 GB
FLUX.2-klein-base-4B3.9BBF164.3 GB
MOSS-Transcribe-Diarize909MBF161.0 GB
trocr-base-printed333MF320.4 GB
kosmos-2-patch14-2241.7BF321.9 GB
flux-3-image-klein5.5BBF166.2 GB
gemma-4-26B-A4B-it-assistant420MBF160.5 GB
trocr-base-handwritten333MF320.4 GB
NuExtract34.5BBF165.1 GB
LightOnOCR-1B-10251.2BBF161.3 GB
mxbai-rerank-large-v21.5BF161.7 GB
GLiNER2.5-Decide486MF320.5 GB
dinov3-vith16plus-pretrain-lvd1689m841MF320.9 GB
gemma-4-12B-it-qat-q4_0-unquantized-assistant423MBF160.5 GB
sundial-base-128m128MF320.1 GB
mxbai-rerank-base-v2494MF160.6 GB
gemma-4-12B-it-assistant423MBF160.5 GB
Janus-Pro-1B2.1BBF162.3 GB
Timestral-209M209MF320.2 GB
laya421MF160.5 GB
stable-fast-3d1.0BF321.1 GB
fg-clip-base150MF320.2 GB
LFM3-Omni-2B2.2BBF162.4 GB
texture-fix-vae-for-qwen-image-2.1338MF320.4 GB
gliner2.5-multi-v1287MF320.3 GB
pixai-tagger-v1.0486MF320.5 GB
GLiNER2.5-multi-Decide287MF320.3 GB

Fits at INT4

Models that fit only after quantizing to 4 bits (0.5 byte per parameter). Needs a quantized checkpoint actually published for the model; check its Hugging Face page before relying on a row.

Text models

The 40 most downloaded of 548.

ModelParametersPublished asVRAM needed
Qwen3-0.6B752MBF160.4 GB
gpt2137MF320.1 GB
Qwen3-8B8.2BBF164.6 GB
Qwen2.5-7B-Instruct7.6BBF164.3 GB
Qwen2.5-1.5B-Instruct1.5BBF160.9 GB
Qwen2.5-3B-Instruct3.1BBF161.7 GB
Llama-3.2-1B-Instruct1.2BBF160.7 GB
Qwen3-4B4.0BBF162.2 GB
Qwen2.5-0.5B-Instruct494MBF160.3 GB
Llama-3.1-8B-Instruct8.0BBF164.5 GB
Qwen3-1.7B2.0BBF161.1 GB
pythia-160m213MF160.1 GB
Qwen3-4B-Instruct-25074.0BBF162.2 GB
gemma-3-1b-it1000MBF160.6 GB
Kimi-K3-DSpark2.2BBF161.3 GB
SmolLM2-135M135MBF160.1 GB
Qwen2.5-Coder-7B-Instruct7.6BBF164.3 GB
Bonsai-27B-mlx-1bit1.7BF161.0 GB
Qwen3-1.7B-Base1.7BBF161.0 GB
Qwen3-4B-Base4.0BBF162.2 GB
Qwen2.5-0.5B494MBF160.3 GB
Meta-Llama-3-8B-Instruct8.0BBF164.5 GB
phi-22.8BF161.6 GB
Llama-3.2-3B-Instruct3.2BBF161.8 GB
SmolLM2-135M-Instruct135MBF160.1 GB
gemma-3-270m268MBF160.1 GB
OpenELM-1_1B-Instruct1.1BBF160.6 GB
Llama-3.2-1B1.2BBF160.7 GB
Mistral-7B-Instruct-v0.27.2BBF164.0 GB
PowerMoE-3b3.4BF321.9 GB
Qwen3-4B-Instruct-2507-FP84.4BF8_E4M32.5 GB
granite-4.1-8b8.8BBF164.9 GB
Qwen2.5-1.5B1.5BBF160.9 GB
gpt2-large812MF320.5 GB
DeepSeek-R1-0528-Qwen3-8B8.2BBF164.6 GB
OLMo-2-0425-1B1.5BF320.8 GB
Qwen3-0.6B-Base596MBF160.3 GB
bloomz-560m559MF160.3 GB
Qwen2-1.5B-Instruct1.5BBF160.9 GB
Qwen2-0.5B494MBF160.3 GB

Vision-language models

The 40 most downloaded of 146.

ModelParametersPublished asVRAM needed
Qwen3.5-9B9.7BBF165.4 GB
Qwen3-VL-8B-Instruct8.8BBF164.9 GB
Qwen2.5-VL-7B-Instruct8.3BBF164.6 GB
Qwen3.5-4B4.7BBF162.6 GB
Qwen3-VL-4B-Instruct4.4BBF162.5 GB
Qwen2.5-VL-3B-Instruct3.8BBF162.1 GB
Unlimited-OCR3.3BBF161.9 GB
chandra-ocr-25.3BBF163.0 GB
Qwen3-VL-2B-Instruct2.1BBF161.2 GB
Qwen3.5-2B2.3BBF161.3 GB
Florence-2-base232MF160.1 GB
DeepSeek-OCR3.3BBF161.9 GB
Qwen3.5-0.8B873MBF160.5 GB
Qwen3-VL-8B-Instruct-FP88.8BF8_E4M34.9 GB
llava-1.5-7b-hf7.1BF163.9 GB
GLM-OCR1.3BBF160.7 GB
Qwen2-VL-2B-Instruct2.2BBF161.2 GB
moondream21.9BBF161.1 GB
gemma-3-4b-it4.3BBF162.4 GB
SmolVLM2-500M-Video-Instruct507MF320.3 GB
Qwen3.5-2B-Base2.3BBF161.3 GB
surya-ocr-2686MBF160.4 GB
Qwen2-VL-7B-Instruct8.3BBF164.6 GB
Qwen3.5-9B-AWQ9.7BBF165.4 GB
DeepSeek-OCR-23.4BBF161.9 GB
medgemma-4b-it4.3BBF162.4 GB
Cosmos-Reason2-2B2.4BBF161.4 GB
Rax-4.52.3BBF161.3 GB
gemma-3-12b-it12.2BBF166.8 GB
GOT-OCR2_0716MBF160.4 GB
Phi-3.5-vision-instruct4.1BBF162.3 GB
HunyuanOCR1.1BBF160.6 GB
UI-TARS-1.5-7B8.3BF324.6 GB
Florence-2-large777MF160.4 GB
llava-onevision-qwen2-0.5b-ov-hf894MF160.5 GB
InternVL2-2B2.2BBF161.2 GB
InternVL2-1B938MBF160.5 GB
SmolVLM-256M-Instruct256MBF160.1 GB
MiniCPM-V-4.61.3BBF160.7 GB
llava-v1.6-mistral-7b-hf7.6BF164.2 GB

Image generation models

The 40 most downloaded of 83.

ModelParametersPublished asVRAM needed
stable-diffusion-xl-base-1.02.6BF321.4 GB
stable-diffusion-v1-5860MF320.5 GB
dreamshaper-7860MF320.5 GB
sdxl-turbo2.6BF321.4 GB
FLUX.1-dev11.9BBF166.7 GB
Z-Image-Turbo6.2BF323.4 GB
FLUX.1-schnell11.9BBF166.6 GB
stable-diffusion-v1-4860MF320.5 GB
RealVisXL_V5.02.6BF321.4 GB
sd-turbo866MF320.5 GB
playground-v2.5-1024px-aesthetic2.6BF321.4 GB
Qwen-Image-2.1-viggle-turbo7.1BBF164.0 GB
animagine-xl-4.02.6BF161.4 GB
animagine-xl-3.12.6BF161.4 GB
nova-furry-xl-il-v120-sdxl2.6BF161.4 GB
one-obsession-17-red-sdxl2.6BF161.4 GB
dreamshaper-8860MF320.5 GB
stable-diffusion-3.5-medium2.5BBF161.4 GB
amanatsu-illustrious-v11-sdxl2.6BF161.4 GB
controlnet-union-sdxl-1.01.3BF160.7 GB
LCM_Dreamshaper_v7860MF320.5 GB
diving-illustrious-real-asian-v50-sdxl2.6BF161.4 GB
controlnet-openpose-sdxl-1.01.3BF160.7 GB
noobai-XL-1.12.6BF161.4 GB
stable-diffusion-xl-1.0-inpainting-0.12.6BF321.4 GB
Qwen-Image-2.17.1BBF164.0 GB
janku-v5-nsfw-trained-noobai-rou-wei-illustrious-xl-v50-sdxl2.6BF161.4 GB
Krea-2-Turbo12.8BBF167.2 GB
stable-diffusion-3.5-large8.1BBF164.6 GB
Illustrious-xl-early-release-v02.6BF161.4 GB
Krea-2-Raw12.8BBF167.2 GB
Chroma1-HD8.9BBF165.0 GB
noobai-XL-Vpred-1.02.6BF161.4 GB
obsession-illustriousxl-v10-sdxl2.6BF161.4 GB
animagine-xl-3.02.6BF161.4 GB
FLUX.1-Krea-dev11.9BBF166.7 GB
hassaku-xl-illustrious-v31-sdxl2.6BF161.4 GB
controlnet-scribble-sdxl-1.01.3BF160.7 GB
ideogram-4-fp89.3BF8_E4M35.2 GB
Z-Image6.2BBF163.4 GB

Video generation models

23 models.

ModelParametersPublished asVRAM needed
LTX-Video1.9BF321.1 GB
Wan2.1-T2V-1.3B-Diffusers1.4BF320.8 GB
stable-video-diffusion-img2vid-xt1.5BF320.9 GB
Wan2.2-TI2V-5B-Diffusers5.0BF322.8 GB
Wan2.2-T2V-A14B-Diffusers14.3BF328.0 GB
Wan2.2-I2V-A14B-Diffusers14.3BF328.0 GB
FastWan2.2-TI2V-5B-FullAttn-Diffusers5.0BBF162.8 GB
Wan2.1-T2V-14B-Diffusers14.3BF328.0 GB
TurboWan2.1-T2V-1.3B-Diffusers1.4BBF160.8 GB
stable-video-diffusion-img2vid1.5BF320.9 GB
Wan2.2-I2V-A14B-Lightning-Diffusers14.3BBF168.0 GB
i2vgen-xl1.4BF320.8 GB
Wan2.1-T2V-14B14.3BF328.0 GB
Wan2.1-T2V-1.3B1.4BF320.8 GB
CogVideoX-2b1.7BF160.9 GB
text-to-video-ms-1.7b1.4BF320.8 GB
LTX-Video-0.9.51.9BBF161.1 GB
CogVideoX-5b5.6BBF163.1 GB
LongLive-2.0-5B-Diffusers5.0BBF162.8 GB
CogVideoX-5b-I2V5.6BBF163.1 GB
Wan2.1-VACE-1.3B-diffusers2.2BF321.2 GB
stable-virtual-camera1.3BF320.7 GB
mochi-1-preview10.0BF325.6 GB

Speech models

The 40 most downloaded of 143.

Embedding models

17 models.

ModelParametersPublished asVRAM needed
Qwen3-Embedding-0.6B596MBF160.3 GB
Qwen3-Embedding-4B4.0BBF162.2 GB
Qwen3-Embedding-8B7.6BBF164.2 GB
multilingual-e5-large-instruct560MF160.3 GB
Qwen3-VL-Embedding-2B2.1BBF161.2 GB
Qwen3-VL-Embedding-8B8.1BBF164.6 GB
gte-Qwen2-1.5B-instruct1.8BF321.0 GB
embeddinggemma-2744MBF160.4 GB
Qwen3-VL-Embedding-8B-FP88.8BF8_E4M34.9 GB
voyage-4-nano346MBF160.2 GB
gte-Qwen2-7B-instruct7.6BF324.3 GB
jina-code-embeddings-0.5b494MBF160.3 GB
stella_en_1.5B_v51.5BF320.9 GB
jina-code-embeddings-1.5b1.5BBF160.9 GB
WeMM-Embedding-2B2.7BBF161.5 GB
WeMM-Embedding-9B9.4BBF165.3 GB
pplx-embed-v2-context-9b-preview8.4BF324.7 GB

Other models

The 40 most downloaded of 88.

ModelParametersPublished asVRAM needed
bert-base-uncased110MF320.1 GB
gemma-4-E4B-it8.0BBF164.5 GB
gemma-4-E2B-it5.1BBF162.9 GB
gemma-4-12B-it12.0BBF166.7 GB
Qwen3-Reranker-4B4.0BBF162.2 GB
sam3860MF320.5 GB
Qwen3-VL-Reranker-2B2.1BBF161.2 GB
Qwen3-Reranker-0.6B596MBF160.3 GB
turn-detector135MF320.1 GB
dinov3-vitl16-pretrain-lvd1689m303MF320.2 GB
gemma-4-E4B-it-qat-w4a16-ct8.7BBF164.9 GB
FLUX.2-klein-4B3.9BBF162.2 GB
gemma-4-E4B8.0BBF164.5 GB
gemma-4-12B-it-FP8-Dynamic13.0BF8_E4M37.2 GB
gemma-4-31B-it-assistant470MBF160.3 GB
blip-image-captioning-large470MF320.3 GB
gemma-4-E2B-it-qat-w4a16-ct5.6BBF163.1 GB
gemma-4-E4B-it-uncensored-heretic8.0BBF164.5 GB
NuMarkdown-8B-Thinking8.3BBF164.6 GB
FLUX.1-Kontext-dev11.9BBF166.7 GB
JEV-9B9.0BBF165.0 GB
bge-reranker-v2-gemma2.5BF321.4 GB
Qwen3.5-4B-3D-Coder4.5BF162.5 GB
FLUX.2-klein-base-4B3.9BBF162.2 GB
FLUX.2-klein-9B9.1BBF165.1 GB
MOSS-Transcribe-Diarize909MBF160.5 GB
gemma-4-12B-it-qat-q4_0-unquantized12.0BBF166.7 GB
trocr-base-printed333MF320.2 GB
kosmos-2-patch14-2241.7BF320.9 GB
flux-3-image-klein5.5BBF163.1 GB
gemma-4-E4B-it-AWQ-INT48.2BF164.6 GB
gemma-4-26B-A4B-it-assistant420MBF160.2 GB
personaplex-7b-v18.4BBF164.7 GB
trocr-base-handwritten333MF320.2 GB
bge-reranker-v2.5-gemma2-lightweight9.2BF325.2 GB
Qwen3-Reranker-8B8.2BBF164.6 GB
gemma-4-12B12.0BBF166.7 GB
NuExtract34.5BBF162.5 GB
openvla-7b7.5BBF164.2 GB
LightOnOCR-1B-10251.2BBF160.6 GB

How these numbers are computed

Required VRAM is the weight size at each precision times a flat 1.2 overhead, the same figure every model page shows. It does not include a long context: the KV cache grows with every token, so a model near the top of a tier can need the next one at long context. Read how much VRAM you need for LLMs for the method, the VRAM and quantization glossary entries for the terms, and the VRAM calculator to size a model that is not listed.

Looking for a model by job rather than by memory? Start with coding, reasoning, chat and assistants, vision-language or see the full models directory.

Other VRAM tiers

Start ComfyUI on a rented GPU.

© 2026 Aquanode. All rights reserved.

All trademarks, logos and brand names are the property of their respective owners.