Self-host an open-weight LLM on a GPU pod
Self-hosting means renting your own GPU and running the model yourself, instead of paying per token for a hosted API. Aquanode sells GPU pods billed per second with no hosted inference API of our own, so this is the actual path every model below takes: size the GPU, install an engine (vLLM, SGLang, Ollama or a GGUF quantization), and run it.
Step 1
Pick a model
Choose an open-weight model whose license fits your use (commercial vs. non-commercial, gated vs. ungated). Every model below links to its own specs page.
Step 2
Size the GPU
Weight size on disk at a given precision (BF16, FP8, INT4) plus a fixed overhead for KV-cache and activations tells you the minimum VRAM. Each deploy guide below computes this from the model's real parameter count, not a guess.
Step 3
Install an inference engine
vLLM and SGLang are the two most common OpenAI-compatible serving engines for LLMs; Ollama and GGUF quantizations (via llama.cpp) trade some throughput for a much smaller memory footprint. Image and video models typically run through Hugging Face's diffusers library or ComfyUI instead.
Step 4
Rent a GPU pod and run it
Aquanode sells GPU pods billed per second, not a hosted inference API: you install and run the engine yourself. A startup script lets you save the install/launch command once and reuse it across pods.
LLM
Deploy Llama 3.3 70B Instruct
vLLM, Ollama, GGUF quantizations
Deploy Qwen2.5 7B Instruct
vLLM, Ollama, GGUF quantizations
Deploy DeepSeek-V3
vLLM, GGUF quantizations
Deploy DeepSeek-V3-0324
vLLM, GGUF quantizations
Deploy DeepSeek-V3.1
vLLM, Ollama, GGUF quantizations
Deploy DeepSeek-V3.2-Exp
SGLang
Deploy EXAONE-3.5-32B-Instruct
vLLM, Ollama, GGUF quantizations
Deploy GLM-4.5-Air
vLLM, GGUF quantizations
Deploy GLM-4.6
vLLM, GGUF quantizations
Deploy GLM-4.7-Flash
vLLM, Ollama, GGUF quantizations
Deploy GLM-5
vLLM
Deploy GLM-5.1
vLLM, GGUF quantizations
Deploy GLM-5.2
vLLM, Ollama, GGUF quantizations
Deploy granite-3.2-8b-instruct
vLLM, Ollama, GGUF quantizations
Deploy Hermes-3-Llama-3.1-405B-FP8
vLLM
Deploy Hermes-3-Llama-3.1-70B
vLLM, GGUF quantizations
Deploy Hermes-3-Llama-3.1-8B
vLLM, GGUF quantizations
Deploy Kimi-K2-Instruct
vLLM, GGUF quantizations
Deploy Kimi-K2-Instruct-0905
vLLM, GGUF quantizations
Deploy Llama-3.1-405B-Instruct
vLLM, Ollama
Deploy Meta-Llama-3.1-70B-Instruct-FP8
vLLM
Deploy Llama-3.1-8B-Instruct
vLLM, Ollama
Deploy Llama-3.1-8B-Lexi-Uncensored-V2
vLLM, GGUF quantizations
Deploy Llama-3.1-Nemotron-Nano-8B-v1
vLLM, GGUF quantizations
Deploy Llama-3.2-3B-Instruct
vLLM, Ollama, GGUF quantizations
Deploy Llama-Guard-3-8B
vLLM, Ollama, GGUF quantizations
Deploy Llama-Guard-4-12B
vLLM
Deploy MiniMax-H3
SGLang, GGUF quantizations
Deploy MiniMax-M2.5
vLLM, GGUF quantizations
Deploy MiniMax-M2.7
vLLM, GGUF quantizations
Deploy Mistral-Small-24B-Instruct-2501
vLLM, GGUF quantizations
Deploy Mixtral-8x7B-Instruct-v0.1
vLLM, GGUF quantizations
Deploy Nous-Hermes-2-Mistral-7B-DPO
vLLM, GGUF quantizations
Deploy Phi-3.5-mini-instruct
vLLM, GGUF quantizations
Deploy Phi-3-mini-4k-instruct
vLLM, GGUF quantizations
Deploy phi-4
vLLM, Ollama, GGUF quantizations
Deploy Qwen2.5-14B-Instruct
vLLM, Ollama, GGUF quantizations
Deploy Qwen2.5-3B-Instruct
vLLM, Ollama, GGUF quantizations
Deploy Qwen3-1.7B
vLLM, Ollama, GGUF quantizations
Deploy Qwen3-1.7B-Base
vLLM, Ollama, GGUF quantizations
Deploy Qwen3-14B-Base
vLLM, Ollama, GGUF quantizations
Deploy Qwen3-235B-A22B-Instruct-2507-FP8
vLLM
Deploy Qwen3-30B-A3B
vLLM, Ollama, GGUF quantizations
Deploy Qwen3-32B
vLLM, Ollama
Deploy Qwen3-4B
vLLM, Ollama
Deploy Qwen3-4B-Base
vLLM, Ollama, GGUF quantizations
Deploy Qwen3.5-122B-A10B
vLLM, Ollama, GGUF quantizations
Deploy Qwen3.5-397B-A17B
vLLM, Ollama, GGUF quantizations
Deploy Qwen3.5-9B
vLLM, Ollama, GGUF quantizations
Deploy Qwen3.6-35B-A3B
vLLM, Ollama, GGUF quantizations
Deploy Qwen3.8-27B-FP8
vLLM, Ollama
Deploy Qwen3-8B
vLLM, Ollama, GGUF quantizations
Deploy ReaderLM-v2
vLLM, GGUF quantizations
Deploy SmolLM2-360M-Instruct
vLLM, GGUF quantizations
Deploy Trinity-Large-Preview
vLLM, GGUF quantizations
Deploy Wayfarer-12B
vLLM, GGUF quantizations
Deploy NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16
vLLM
Reasoning
Deploy DeepSeek-R1-0528
vLLM, GGUF quantizations
Deploy DeepSeek-R1-Distill-Llama-70B
vLLM, GGUF quantizations
Deploy DeepSeek-R1-Distill-Llama-8B-abliterated
vLLM, GGUF quantizations
Deploy DeepSeek-R1-Distill-Qwen-1.5B
vLLM, GGUF quantizations
Deploy DeepSeek-R1-Distill-Qwen-14B-abliterated-v2
vLLM, GGUF quantizations
Deploy Dolphin3.0-R1-Mistral-24B
vLLM, GGUF quantizations
Deploy Qwen3-235B-A22B-Thinking-2507
vLLM, GGUF quantizations
Deploy Qwen3-Next-80B-A3B-Instruct
vLLM, Ollama, GGUF quantizations
Deploy Qwen3-Next-80B-A3B-Thinking
vLLM, Ollama, GGUF quantizations
Deploy gpt-oss-120b
vLLM, Ollama, GGUF quantizations
Deploy gpt-oss-20b
vLLM, Ollama, GGUF quantizations
Coding
Vision
Deploy DeepSeek-OCR
vLLM, Ollama
Deploy gemma-3-12b-it
vLLM, Ollama, GGUF quantizations
Deploy gemma-3-4b-it
vLLM, Ollama, GGUF quantizations
Deploy gemma-4-31B-it
vLLM, Ollama, GGUF quantizations
Deploy GLM-4.5V
vLLM, GGUF quantizations
Deploy InternVL3-78B
vLLM, GGUF quantizations
Deploy Llama-4-Maverick-17B-128E-Instruct-FP8
vLLM
Deploy Llama-4-Scout-17B-16E-Instruct
vLLM, GGUF quantizations
Deploy Nanonets-OCR-s
vLLM, GGUF quantizations
Deploy Qwen3-VL-32B-Instruct
vLLM, Ollama, GGUF quantizations
Image generation
Deploy FLUX.1 [dev]
Diffusers (Python), ComfyUI
Deploy FLUX.1-Kontext-dev
Diffusers (Python), ComfyUI
Deploy FLUX.1-schnell
Diffusers (Python), ComfyUI
Deploy FLUX.2-dev
Diffusers (Python), ComfyUI
Deploy HiDream-I1-Full
Diffusers (Python), ComfyUI
Deploy Krea-2-Raw
Diffusers (Python), ComfyUI
Deploy Qwen-Image
Diffusers (Python), ComfyUI
Deploy RealVisXL_V5.0
Diffusers (Python), ComfyUI
Deploy stable-diffusion-xl-base-1.0
Diffusers (Python), ComfyUI
