Pricing
Coming soon.
Caldrix is free on your own machine today. Paid plans aren't ready yet.
Inference providers
Run faster on the hardware you already own.
We work directly with inference providers, under contract, to compile for the fleet they have.
Your fleet
We compile for the machines you run: AMD GPUs, Apple Silicon and x86 CPUs today, other silicon as each lane lands.
Your models and your load
The build takes your model, context window and concurrency as inputs, so it is built for your traffic, not for an average machine.
Measured on your machines
You run the comparison on your own hardware. We claim nothing we have not measured there.
Contact
Tell us what you run.
A few lines about your hardware and your models is enough. We reply by email.
The form opens shortly.
Get started
Start free on your own machine.
Install the CLI and run a model. Talk to us when you want it faster on your fleet.