Hybrid · Bring your own GPUs
Your GPUs. Our platform.
Already have the hardware? Plug your GPUs into Yantra as a private, pinned fleet. Intelligaia runs the control plane — routing, governance, logging — while your compute stays yours. You pay a platform fee plus a small managed fee per GPU, not per token.
How it's priced
A platform fee, then per GPU.
No usage metering — you own the compute, so you run it as hard as you like. Add a support tier for managed operations.
The Yantra gateway managing your fleet: SSO, roles, token approval, routing & failover, logging, and the full model registry.
Each GPU you attach is monitored, kept warm, tuned, and pinned. Add or remove nodes and the fee tracks your fleet.
Integrate your nodes, wire up identity, deploy the workers, and validate every modality against your hardware.
Worked example
An 8-GPU production fleet.
Plus a one-time $18,000 onboarding. Drop to Standard support and the monthly is $11,600. No per-request charges — run unlimited inference on your own silicon.
The split
What you own, what we run.
You own
- The GPUs and the data center they sit in
- All model weights and data — nothing leaves
- Power, cooling, and hardware lifecycle
- Unlimited inference — no metering
We run
- The control plane: routing, auth, logging
- Node monitoring, warm-up, and GPU tuning
- Upgrades, patches, and model onboarding
- Incident response to your support SLA
Put your GPUs to work
Let's size your fleet.
Tell us how many GPUs you have and the modalities you need, and we'll return a monthly run-rate.