Own US-based AI infrastructure

Buy an NVIDIA-powered GPU server you own. We build it by hand in Eugene, Oregon and host it; you choose how it runs, and you keep 85% of what it earns.

How it works

  1. You configure and buy it. Pick the GPU, memory and storage, and whether it racks with us or ships to you. It is yours outright.
  2. We build and host it. Hand-built in Eugene, Oregon and racked in our facility: 24/7 power, cooling, and on-site engineering.
  3. You choose how it runs. Earning, under the offtake contracts we hold, and you keep 85% of what it earns; or private, running your workloads alone for a flat monthly fee.

You hold title, and you have root on the machine. Monetizing is an option, not an obligation.

Earnings depend on marketplace demand: projections, not guarantees.

Machines

Own it, don't rent it

Cloud rent is a metered bill that never ends and buys you nothing. A machine you own stays yours, and earns whenever you make it available. Hosting it with us costs a fraction of that rent, and on an earning machine it comes out of what the machine makes rather than out of your pocket.

There's a tax difference too: cloud rent is 100% expense, forever. A machine you own is a depreciable business asset. How that applies to you depends on your situation; consult your tax advisor.

Your models run on hardware you own in a US facility you can name: weights, keys, and data never leave machines you control. Hand-built GPU clusters, US-assembled by Andromeda, with N+1 redundant power. Hosted GPUs for Stanford Blockchain, Penn Blockchain, NYU, and Dartmouth Blockchain.

Browse machines · Financing · Docs · Book a 30-minute call

What hardware runs each model

For every model B3IQ routes: the cheapest machine that serves it, the cheapest that fine-tunes it, and the throughput real operators report.

ModelParamsServe VRAMCheapest machine
Qwen3 8B (llama.cpp)8Bnot listedNot on a machine we sell
Qwen3 14B (llama.cpp)14Bnot listedNot on a machine we sell
Qwen3 8B8Bnot listedNot on a machine we sell
Mistral 7B7Bnot listedNot on a machine we sell
Granite 3.3 8B8Bnot listedNot on a machine we sell
Private Chat 8B8B8 GB1x RTX 5090
Qwen Coder 7B7B8 GB1x RTX 5090
DeepSeek-R1 7B7B8 GB1x RTX 5090
DeepSeek-R1 8B8B8 GB1x RTX 5090
Code Workbench 14B14B16 GB1x RTX 5090
Qwen3 14B14B16 GB1x RTX 5090
Phi-4 14B14B16 GB1x RTX 5090
DeepSeek-R1 14B14B16 GB1x RTX 5090
Gemma 3 12B12B16 GB1x RTX 5090
Gemma 3 27B27B18 GB1x RTX 5090
gpt-oss 20B20B16 GB1x RTX 5090
Mistral Small 24B24B16 GB1x RTX 5090
vLLM Qwen3 14B14B24 GB1x RTX 5090
vLLM MiniMax M2230B138 GB2x RTX PRO 6000
Qwen Quality 32B32B22 GB1x RTX 5090
Qwen3 30B A3B30B20 GB1x RTX 5090
Qwen3 30B A3B30B32 GB1x RTX 5090
Qwen3 Coder 30B30B20 GB1x RTX 5090
Qwen Coder 32B32B32 GB1x RTX 5090
DeepSeek-R1 32B32B32 GB1x RTX 5090
Llama 3.3 70B70B42 GB2x RTX 5090
gpt-oss 120B120B72 GB1x RTX PRO 6000
Qwen3 Coder 8B8Bnot listedNot on a machine we sell
Devstral Small 24B24B16 GB1x RTX 5090
Qwen3 Coder 27B27B32 GB1x RTX 5090
DeepSeek V4 Pro671B400 GB4x H200 NVL
Kimi K2.6 MoE1000B600 GB8x H200 NVL
Qwen2.5 7B Instruct7B8 GB1x RTX 5090
Qwen2.5 14B Instruct14B16 GB1x RTX 5090
Qwen2.5 32B Instruct32B32 GB1x RTX 5090
Llama 3.2 Vision 11B11B8 GB1x RTX 5090
LLaVA 7B (Vision)7Bnot listedNot on a machine we sell
Qwen3 Embedding 8B8B18 GB1x RTX 5090
gte-Qwen2 7B7B16 GB1x RTX 5090
Qwen2.5-VL 7B7B7 GB1x RTX 5090
Qwen2.5-VL 32B32B24 GB1x RTX 5090
Qwen2.5-VL 72B72B45 GB2x RTX 5090
MiniCPM-V 2.6 8B8B7 GB1x RTX 5090
LLaVA-Llama3 8B8B7 GB1x RTX 5090
LLaVA 13B13B10 GB1x RTX 5090
LLaVA 34B34B24 GB1x RTX 5090
BakLLaVA 7B7B6 GB1x RTX 5090
Mistral Small 3.1 24B (Vision)24B18 GB1x RTX 5090
Llama 3.2 Vision 90B90B55 GB2x RTX 5090
Wan2.1 14B (Xinference)14B40 GB2x RTX 5090
HunyuanVideo (Xinference)13B48 GB2x RTX 5090