HomePricingBlogsDocs

Open-weight AI models you can run on a GPU pod

VRAM requirements and the cheapest live GPU fit for 1,461 Hugging Face models, grouped by model line and publisher, at native, FP8 and INT4 precision. Aquanode doesn't run a hosted inference API: every model page links to a GPU pod sized for it, billed per second, that you install your own inference stack on (or, for image and video models, our ComfyUI template). Not sure which model to run in the first place? See how the leading open-weight families score on published benchmarks in the LLM benchmarks leaderboard.

Model series

Every generation of a model line on one page, with the sizes, context lengths and VRAM of each.

Qwen

Alibaba (Qwen) · 8 generations · 311 models

Llama

Meta · 7 generations · 96 models

Gemma

Google · 5 generations · 79 models

DeepSeek

DeepSeek · 3 generations · 42 models

Mistral

Mistral AI · 4 generations · 14 models

GLM

Zhipu AI (Z.ai) · 3 generations · 39 models

Kimi

Moonshot AI · 2 generations · 9 models

MiniMax

MiniMax · 3 generations · 16 models

Hunyuan

Tencent · 3 generations · 10 models

Phi

Microsoft · 5 generations · 21 models

Granite

IBM · 2 generations · 38 models

Nemotron

NVIDIA · 1 generation · 35 models

OLMo

Ai2 · 3 generations · 24 models

SmolLM

Hugging Face · 3 generations · 14 models

Yi

01.AI · 2 generations · 11 models

Falcon

TII · 2 generations · 14 models

FLUX

Black Forest Labs · 2 generations · 11 models

Wan

Alibaba (Wan) · 2 generations · 18 models

Stable Diffusion

Stability AI · 4 generations · 47 models

InternVL

Shanghai AI Lab (OpenGVLab) · 2 generations · 8 models

Best models by task

Ranked lists with the VRAM and live GPU fit of each model.

Best coding models

32 models

Best reasoning models

29 models

Best chat and assistants models

480 models

Best vision-language models

153 models

Best OCR and document parsing models

27 models

Best speech-to-text models

52 models

Best text-to-speech models

23 models

Best image generation models

44 models

Best video generation models

31 models

Best embeddings models

10 models

Best translation models

9 models

Models by GPU memory

Find what fits the GPU you have or plan to rent.

All model families

Every model line, grouped by publisher.

Qwen

Qwen3

103 models · 460M to 480.2B

Qwen2.5

47 models · 135M to 73.4B

Qwen3.5

56 models · 556M to 403.4B

Qwen3.6

17 models · 386M to 36.0B

Qwen3.8

42 models · 1.9B to 2446.2B

Qwen2

15 models · 494M to 72.7B

Qwen 1

3 models · 7.7B to 72.3B

Qwen-Image

10 models · 338M to 27.4B

Qwen1.5

18 models · 464M to 72.3B

Meta

Llama 3.2

15 models · 506M to 88.6B

Llama 3.1

40 models · 950M to 405.9B

Llama 3

19 models · 8.0B to 70.6B

Llama 2

8 models · 6.7B to 13.0B

Llama 3.3

4 models · 49.9B to 70.6B

Code Llama

3 models · 6.7B to 33.7B

Llama Guard

4 models · 1.2B to 12.0B

Google

Gemma 4

42 models · 420M to 32.7B

Gemma 3

11 models · 268M to 27.4B

Gemma 2

11 models · 2.5B to 27.2B

google

3 models · 105M to 25.8B

MedGemma

4 models · 4.3B to 28.8B

PaliGemma

3 models · 2.9B to 2.9B

Gemma 3n

4 models · 5.4B to 8.4B

Gemma 1

3 models · 2.5B to 8.5B

Stability AI

Stable Diffusion XL

34 models · 611M to 2.6B

Stable Diffusion 1

3 models · 860M to 860M

Stable Diffusion 2

5 models · 866M to 866M

Stable Diffusion 3

3 models · 2.1B to 8.1B

stabilityai

4 models · 568M to 1.3B

StableLM

3 models · 1.6B to 2.8B

NVIDIA

Parakeet

7 models · 609M to 1.1B

Nemotron 3

18 models · 638M to 560.5B

Cosmos

6 models · 2.4B to 64.6B

Nemotron Nano

5 models · 4.0B to 13.2B

Nemotron Labs

5 models · 3.8B to 13.5B

Nemotron-H

3 models · 8.1B to 56.3B

Z.ai

GLM-5

23 models · 1.2B to 753.9B

GLM-4.5

10 models · 10.3B to 358.5B

GLM-4

6 models · 9.4B to 32.6B

CogVideoX

3 models · 1.7B to 5.6B

DeepSeek

DeepSeek R1

14 models · 1.8B to 684.5B

DeepSeek V3

7 models · 684.5B to 685.4B

DeepSeek Coder

6 models · 1.3B to 15.7B

DeepSeek V2

5 models · 15.7B to 235.7B

DeepSeek VL

3 models · 2.0B to 7.3B

DeepSeek V4

3 models · 304.2B to 1650.5B

Meta

Wav2Vec 2.0

32 models · 315M to 963M

facebook

4 models · 303M to 2.3B

MMS

3 models · 315M to 965M

IBM

Granite 4

20 models · 352M to 32.2B

Granite Speech

7 models · 473M to 8.6B

Granite 3

9 models · 1.3B to 8.2B

Microsoft

Florence-2

5 models · 232M to 777M

Phi-4

7 models · 3.8B to 14.7B

Phi-3.5

4 models · 3.8B to 41.9B

Phi-3

6 models · 3.8B to 14.0B

VibeVoice

4 models · 323M to 8.7B

microsoft

7 models · 176M to 4.7B

OpenAI

Whisper

30 models · 242M to 1.6B

gpt-oss

6 models · 350M to 116.8B

Ai2

OLMo 2

5 models · 1.4B to 32.2B

Olmo 3

8 models · 7.3B to 32.2B

OLMoE

4 models · 6.9B to 6.9B

Molmo

3 models · 4.9B to 8.7B

EleutherAI

Pythia

15 models · 213M to 12.0B

GPT-Neo

3 models · 150M to 2.7B

Liquid AI

LFM2

15 models · 230M to 8.5B

LiquidAI

5 models · 1.7B to 2.2B

Hugging Face

SmolLM2

6 models · 135M to 1.7B

SmolVLM

4 models · 256M to 2.2B

SmolLM3

3 models · 3.1B to 3.1B

SmolLM-1

5 models · 135M to 1.7B

Wan-AI

Wan2.1

11 models · 1.4B to 17.3B

Wan2.2

7 models · 5.0B to 16.3B

ornith-ai

Ornith 1

17 models · 4.3B to 403.4B

MiniMax

MiniMax H3

6 models · 27.4B to 35.0B

MiniMax M2

4 models · 228.7B to 228.7B

MiniMax M3

4 models · 3.1B to 440.3B

TII

Falcon 1

4 models · 6.9B to 41.8B

Falcon-H1

4 models · 521M to 7.6B

Falcon3

4 models · 1.7B to 7.5B

Mistral AI

Mistral 7B

7 models · 7.2B to 7.2B

Mistral Small

3 models · 23.6B to 24.0B

Tencent

tencent

3 models · 1.1B to 4.8B

Hy3

3 models · 298.8B to 298.8B

Hy-MT2

3 models · 2.0B to 30.1B

Black Forest Labs

FLUX.1

5 models · 347M to 11.9B

FLUX.2

6 models · 3.9B to 32.2B

Moonshot AI

Kimi K3

3 models · 2.2B to 3.6B

Kimi K2

5 models · 3.0B to 1026.9B

OpenBMB

MiniCPM-V

3 models · 1.3B to 8.7B

MiniCPM5

3 models · 1.1B to 1.1B

MiniCPM4

4 models · 434M to 8.2B

01.AI

Yi-1.5

6 models · 6.1B to 34.4B

Yi-1

5 models · 6.1B to 34.4B

inclusionAI

LLaDA

5 models · 8.0B to 16.3B

Ling

3 models · 7.9B to 127.5B

llava-hf

LLaVA

9 models · 894M to 13.4B

openai-community

GPT-2

8 models · 110M to 1.6B

LG AI Research

EXAONE 3.5

4 models · 2.4B to 32.0B

EXAONE 4.0

3 models · 1.3B to 32.0B

OpenGVLab

InternVL2

4 models · 938M to 25.5B

InternVL3

4 models · 938M to 78.4B

Baidu

baidu

4 models · 3.3B to 286.2B

ERNIE 4.5

3 models · 361M to 21.9B

internlm

InternLM2

5 models · 7.7B to 19.9B

Nous Research

Hermes 3

3 models · 8.0B to 405.9B

Hermes 2

4 models · 6.7B to 46.7B

ATH-MaaS

Ovis

6 models · 853M to 10.2B

bigscience

BLOOM

6 models · 559M to 176.2B

Cohere

CohereLabs

3 models · 2.1B to 30.5B

typhoon-ai

Typhoon

6 models · 2.1B to 30.5B

upstage

SOLAR

6 models · 10.7B to 250.3B

Xiaomi MiMo

MiMo V2

4 models · 9.4B to 1023.2B

Lightricks

LTX

5 models · 1.9B to 19.0B

Salesforce

BLIP

3 models · 470M to 3.9B

arcee-ai

Trinity

3 models · 6.1B to 398.6B

KBlueLeaf

KBlueLeaf

4 models · 393M to 508M

meituan-longcat

LongCat

4 models · 69.1B to 1779.8B

Nanbeige

Nanbeige

4 models · 848M to 33.1B

parler-tts

Parler-TTS

4 models · 878M to 2.3B

poolside

Laguna

4 models · 33.4B to 117.6B

sensenova

sensenova

4 models · 9.7B to 27.4B

StepFun

stepfun-ai

4 models · 561M to 321.0B

Zyphra

Zamba

4 models · 1.2B to 7.4B

ai21labs

Jamba

3 models · 3.0B to 51.6B

briaai

briaai

3 models · 8.1B to 8.3B

ByteDance

ByteDance

3 models · 1.4B to 2.7B

ByteDance Seed

ByteDance-Seed

3 models · 8.3B to 36.2B

convaiinnovations

convaiinnovations

3 models · 322M to 421M

datalab-to

datalab-to

3 models · 686M to 9.7B

fastino

GLiNER

3 models · 287M to 486M

google-t5

T5

3 models · 223M to 2.9B

h2oai

h2oai

3 models · 514M to 2.2B

HuggingFaceM4

Idefics

3 models · 8.4B to 8.9B

kakaocorp

kakaocorp

3 models · 1.3B to 30.7B

lightonai

lightonai

3 models · 1.0B to 1.2B

naver-hyperclovax

naver-hyperclovax

3 models · 3.7B to 33.3B

OpenMOSS-Team

OpenMOSS-Team

3 models · 909M to 5.1B

pfnet

pfnet

3 models · 890M to 31.8B

ricursiveAI

ricursiveAI

3 models · 4.0B to 64.8B

skt

skt

3 models · 7.3B to 691.7B

state-spaces

Mamba

3 models · 129M to 2.8B

swiss-ai

Apertus

3 models · 8.1B to 70.6B

utter-project

EuroLLM

3 models · 1.7B to 9.2B

vikhyatk

Moondream

3 models · 1.9B to 9.3B

XHToken

XHToken

3 models · 1.7B to 4.1B

More models

Single models and small groups of models from one publisher.

Search all models

Find any of the 1,461 models by name.

How the VRAM and cost numbers are calculated

  • Weight bytes = parameter count × bytes-per-parameter for the shown precision (FP32=4, BF16/FP16=2, FP8=1, INT4=0.5 bytes).
  • Required VRAM is the weight size times a flat 1.2× factor for activations and allocator fragmentation. That factor is a rule of thumb, not a per-request simulation, and it does not contain the KV cache. Diffusion and video models scale mainly with output resolution and frame count, and speech models mainly with input audio length, neither of which it models.
  • The KV cache is a separate, per-model table on each language-model page and in the “Pick a size” table on each hub. It is computed from the layer, head and window counts in the model's own configuration, at 16 bits per cached value, for one sequence, and grows with context length and with every concurrent request. Where the architecture is not published the page says so and shows no number. Add it to the required VRAM above when you plan a long context or a batch.
  • “Cheapest live fit” only matches a GPU to a row if its hardware actually supports that row's compute format, and caps a fit at 8 GPUs of the same model. A pre-Ampere card (Pascal/Volta/Turing) never qualifies for a BF16 or FP8 row, regardless of VRAM or price. See the GPU Index for per-model specs.
  • The INT4 row additionally requires a quantized checkpoint that's actually been published for the model in question. For language models this also means a GPU with AWQ/GPTQ/Marlin-class quantized-serving kernel support. Neither is guaranteed, so check the model's Hugging Face page for a published quantized variant before relying on that row.
  • Prices are live per-GPU $/hr rates from Aquanode's live price feed, normalized the same way as the GPU Index, with outlier offers excluded.
  • Multi-GPU fits assume the model shards cleanly across cards (e.g. tensor parallelism). Real throughput and interconnect overhead aren't modeled.
VRAM calculatorFine-tune a modelPrototype in a notebookBrowse the GPU Index

Submit the job. Everything after that is ours.

Sign up in 60 seconds. Pay for the GPU minutes you actually use.

Aquanode LogoAquanode

Run AI without spending any time on infrastructure.

© 2026 Aquanode. All rights reserved.

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