Two offerings #
Dedicated hosting changes where an organisation’s API requests run; Web Hosting gives it a static site. Each is requested on its own, and the two combine.
Compute
Dedicated hosting
From 249 CAD a month
Private, always-warm capacity for one organisation: a private Falcon API and Express Cue, plus one or two NVIDIA L4 GPUs on the GPU tiers.
Add-on
Web Hosting
9 CAD a month
One static site per organisation at https://<site>.falconlab.app with TLS and a global CDN, uploaded as a zip from the portal. Combines with any tier.
Which one do you need #
| At a glance | Dedicated hosting | Web Hosting |
|---|---|---|
| What it is | Private Falcon API and Express Cue instances kept warm for one organisation, plus private L4 GPUs for the GPU models on the GPU tiers | One static site: HTML, CSS, JavaScript, images and fonts |
| Who it is for | Integrators whose traffic cannot wait for a GPU service to wake up, or that want their own always-warm instances rather than the shared ones | Organisations that want a landing page, documentation or a demo next to their integration |
| Where it runs | us-central1, operated by Ducky Software | A global CDN, over TLS |
| What changes for you | The same keys, routes and responses. Express Cue and the hosted GPU models switch over on their own; the in-process models run privately at a private API address you call | A site at https://<site>.falconlab.app, or a custom domain on request |
| Lead time | Dedicated Warm one business day after approval; GPU tiers when GPU capacity is available | One business day after each upload |
| Combines with | Web Hosting | The shared tier or any dedicated tier |
How it works #
- Sign in to the portal with the credentials an administrator issued for your organisation.
- Send a request from the dashboard: a tier in the Hosting panel, or a site name in the Web hosting panel.
- An administrator reviews it and provisions Dedicated Warm, or publishes the site, within one business day. A GPU tier waits until GPU capacity is free.
- The panel shows Active with the date. Billing starts that day, monthly in CAD.
Questions #
Do I need dedicated hosting to use Falcon? #
No. Every preview organisation uses the shared fleet, metered by the quota buckets on Access. A tier changes where requests run, not what they can do.
Does my code change? #
Only the base URL, and only if you want the in-process models to run privately. Keys, routes, request and response shapes and error codes stay as on API conventions.
Can I have both? #
Yes. The two are independent; Web Hosting also combines with the shared tier.
Which models can run on dedicated capacity? #
The Falcon API models and Express Cue run on every dedicated tier, and one GPU model service per L4 on the GPU tiers (Waymark and Waymark Extra count as one); What runs where lists them.
How do I cancel? #
A pending request can be cancelled from its panel on the dashboard. An active tier or site is cancelled by asking an administrator, and ends at the end of the current month.
Who can request hosting? #
A signed-in member of an organisation in the private preview. Self-serve accounts are paused.