Hinode — Your AI workstation in the cloud<br>Sign inGet started
Your AI workstation in the cloud<br>A Linux desktop with a dedicated GPU — in your browser or the native Mac and Windows apps. Your files and environment stay put. Boost to a bigger GPU in one click.<br>Create workstationSee pricing<br>From $3.99/hr · Auto-pause when idle · Native Mac & Windows apps
hinode · ai-workstation · L4
07:04
ComfyUI — Image Generation
Image Generation<br>Run
Load Diffusion ModelMODEL<br>z_image_turbo.sft
CLIP Text EncodeCOND<br>golden sunrise over a calm sea, warm cinematic light, 4k
Empty Latent Image<br>1024 × 1024 batch 1
KSamplerLATENT<br>seed 42<br>steps 20 · cfg 6.5<br>euler · karras
VAE DecodeIMAGE<br>samples → image
Save Image
1024 × 1024
Blender — suzanne.blend
User Perspective<br>(1) Suzanne<br>Verts 518 · Faces 999 · Cycles
hinode@l4-station
$ comfy run sunrise_workflow.json<br>[comfy] z_image_turbo loaded · L4 · 24 GB<br>[comfy] sampling 20 steps · 1.9 it/s<br>[comfy] saved → output/sunrise_0042.png
The problem<br>GPU rental wasn't built for daily work<br>Rented pods are disposable. Your work isn't. The same evening, two ways.
Raw GPU pod
pod-7f3a · H100
$ rsync -r checkpoints/ loras/ pod-7f3a:<br>sent 3.2 GB · done<br>$ python generate.py --batch 24<br>done · 24 images · 214 s<br>Terminating instance and cleaning up...
checkpoints/loras/outputs/
Pods tear down when you stop. Your models, nodes and files go with them. The next day, you start over from scratch.
Hinode workstation<br>✓ resumed<br>Files — ~/projects<br>checkpoints/<br>loras/<br>outputs/
ComfyUI — Image Generation
paused · $0.00/hr
Hinode auto-pauses instead. Close the lid — files, apps and processes wait intact. Open it and continue where you left off.
Boost<br>When the model doesn't fit<br>The same job on the same 24 GB card — with and without a way out.
Raw GPU pod
pod-9c21 · L4
$ comfy run flux_portrait.json<br>loading flux-dev · 23.8 GB<br>CUDA out of memory — job killed
VRAML4 · 24 GB
Fixed VRAM is a wall. The model needs more than the card has, and the job dies. Creating a new pod means starting over from scratch.
Hinode workstation
hinode · ai-workstation
$ comfy run flux_portrait.json<br>loading flux-dev · 23.8 GB<br>✓ loaded · 23.8 / 96 GB VRAM
VRAML4 · 24 GBRTX Pro 6000 · 96 GB
⚡ Boosted
Boost raises the ceiling. Boost to the Power tier when you need it, and pay only for the time you use it.
Features<br>One workstation, always ready<br>Everything an AI creator needs in a single environment that follows you across devices.
Persistent state + auto-pause<br>Your workstation hibernates after 15 minutes idle and wakes with everything intact. You don't pay for idle time.
Boost<br>One click swaps in a bigger GPU for a single training run, billed by the hour. It reverts automatically when you're done.
Hybrid billing<br>One product, two ways to pay. Monthly subscription if you work daily, hourly pay-as-you-go if you don't. Switch any time.
H.265 60fps streaming<br>Every tier streams 1440p60 with modern compression. Upgrades buy GPU power, not resolution.
AI-ready image<br>The AI Workstation image ships with CUDA, PyTorch, JupyterLab, VS Code, and the common ML libraries. Open it and start.
Browser or native app<br>Run it in your browser, or get the native Mac and Windows apps — the same live session either way.
No lock-in<br>Your tools, your data<br>A full Linux desktop with the tools you already use, preinstalled. Import your data from any source and export it to any destination. No vendor lock-in, no closed ecosystems.<br>ComfyUIBlenderPyTorchHugging FaceOllamaCUDALangChainvLLMJupyter
Pricing<br>Pay monthly, or by the hour<br>Same workstation either way. Switch anytime.
SubscriptionPay-as-you-go
Standard<br>Dedicated L4<br>$3.99/hr<br>24 GB VRAM<br>32 GB RAM<br>Billed by the active minute<br>Persistent state + auto-pause<br>Get started
Most popularPro<br>Dedicated L40S<br>$4.99/hr<br>48 GB VRAM<br>64 GB RAM<br>Billed by the active minute<br>Persistent state + auto-pause<br>Get started
Power<br>Dedicated RTX Pro 6000<br>$6.99/hr<br>96 GB VRAM<br>128 GB RAM<br>Billed by the active minute<br>Persistent state + auto-pause<br>Get started
What a month actually costs<br>Pay-as-you-go bills active time by the minute — auto-pause stops the meter when you step away.<br>Active work20 h/week<br>≈ 87 active hours a month
Pay-as-you-goSubscription<br>Standard$347$299<br>Pro$434$399<br>Power$608$599
Subscriptions include 100 active hours per workstation. Extra active time uses the lower subscriber rate shown above.
Storage billed separately by allocated GB-minute · Boost billed per hour at the delta rate
How we compare<br>The same GPU. A full desktop.<br>Pro, our most popular tier, is a dedicated L40S — the same card RunPod rents — with a full desktop that stays put instead of a terminal you rebuild.
RunPodVastLambdaHF ProHinode ProGPUL40S · 48 GBRTX A6000 · 48 GBA100 · 40 GBZeroGPU · sharedL40S · 48 GBGPU accessDedicatedDedicatedDedicatedQuota + queueDedicatedInterfaceSSH / JupyterSSH / JupyterSSH / JupyterSpaces (web)Native desktopSessionLost on stopTied to one hostLost on terminateEphemeralPersistent +...