Where it runs
Eigna Hybrid
Small experts on your machine, the main expert on a GPU in your region.
Where it runs
- Small expertsOn your machine
- Main expertGPU in your region
- Prompt logsOff by default
- Region you choose for the main expert
- 1
- Pool tokens a month per site, as planned
- 30M
- To roll back to the last release, by design
- 60 s
- A site a month, proposed
- $1,900
Eigna Hybrid
The spec, as planned.
- Runs on
- Small experts on your machines. The main expert on a GPU in your cloud or ours, in your region.
- Where data stays
- Data stays in the region you choose.Sensitive work stays on your machines. Prompts for the main expert go to our GPU pool in your region over mutual TLS, with no prompt logs by default.
- When it learns
- From approved work only. The training path for Hybrid is confirmed in your offer.
- Keys
- Held by you.
- Best for
- Teams whose machine is too small for the main expert.
- Teams with a desk machine, not a data centre
- Work that is sensitive in parts
- Offices that want local speed for routine tasks
Setup
How it is set up.
We deliver it, set it up and run it. No ML team to hire.
- 01Test LabOne task, about four weeks, on your own data. The test set becomes your first release gate.
- 02Choose the homeDecided by where the data must stay.
- 03InstallSmall experts installed on your machines with the signed installer, the main expert on a GPU in your region. Keys held by you.
- 04LearnApproved work trains the next release. Live only when better.
As planned
Limits and price, as planned.
Setup
$3,250
a site
Running
$1,900
a site a month
Proposed prices, not yet tested with buyers. Every install starts with a Test Lab, $4,900 for four weeks, credited against setup.
- Users
- Up to 15
- Expert scopes
- One, more at $325 a month each
- Pool tokens
- 30 million a month per site; $170 per extra 30 million, only if you opt in
Eigna Hybrid
Sensitive work stays inside.
Small experts answer on your machines. Only prompts for the main expert travel, over mutual TLS, to a GPU in your region.
Questions and answers
What leaves our site?
Prompts for the main expert travel to our GPU pool in your region over mutual TLS, with no prompt logs by default. Sensitive work stays on your machines.
Why not run everything locally?
When your machine is too small for the main expert, Hybrid keeps small experts on site and runs the main expert on a GPU in your region.
Is there a token limit?
As planned, 30 million pool tokens a month per site, with a top-up of $170 per 30 million, only if you opt in.
Start
Not sure which mode fits?
The fit check recommends one in two minutes, from where your data must stay.