Models and sensitivity tiers

Sensitivity tiers

Every request is classified by content sensitivity and routed to a model approved for that classification.

Tier	Content sensitivity
T1	Highest — regulated, restricted
T2	Sensitive — business or personal data
T3	Non-sensitive

Tier decisions are made by the guardrail on each request — see How it works.

Where inference runs

Routing per tier is a per-account configuration. Depending on your policy, tiers may route to Advania-operated GPUs, dedicated hardware, or approved cloud providers. Contact your Advania Solutions Architect to set your routing policy.

Available models

Confirmed catalog for launch — pending sign-off from Model Access team. This page will be replaced by the live model catalog once the API is populated.

Use GET /v1/models or browse to the Model Catalog to list what your account can call today:

curl -sS https://api.alt.advania.ai/v1/models \
  -H "Authorization: Bearer $ALT_API_KEY"

The response follows the OpenAI models shape. Each entry includes the model id you pass in chat.completions.create({ model: … }).