Skip to content

6.11.2.2 AgentModelProfile

6.11.2.2.1 Purpose

A model and its settings as the AI gateway calls it: provider, model id, region, the data classes it may receive, its tier and parameters. Profiles are created, edited and retired from the Console; the choice of model behind an agent is a configuration change with an audit event, never a deploy.

6.11.2.2.2 Questions it answers

  • Which models are available, where do they run, and what data may they see?
  • Which profile does a plan tier map to?
  • What does a profile cost per million tokens, for the usage ledger?

6.11.2.2.3 Fields

Field Type Cardinality Description
id string (uuid) 1..1 Key.
slug string 1..1 Unique.
e.g. gemini-flash-eu, gemini-pro-eu, openrouter-public.
provider enum 1..1 Values: vertex_ai, openrouter.
model string 1..1 The provider's model id, e.g. gemini-3.8-flash.
region string 0..1 e.g. europe-west8; empty for a provider that does not pin a region.
data_classes enum 1..* Values: public, personal.
Which data the profile may receive. personal requires an EU region and zero data retention.
tier enum 1..1 Values: lite, standard, pro.
What a plan's model_tier (see Entitlement, Entitlement.limits) maps to.
parameters string (JSON) 1..1 Temperature, output cap, context cap, thinking budget.
cost_input_micros integer 1..1
cost_output_micros integer 1..1 Provider list price per 1M tokens in millionths of a euro, for the ledger; updated when the provider changes it.
status enum 1..1 Values: active, retired. Default: active.
system_version SystemVersion 1..1
created_at datetime 1..1
updated_at datetime 1..1

6.11.2.2.4 Relationships

Referenced by

Uses types

SystemVersion

6.11.2.2.5 Rules & constraints

  • A model and its settings, as the AI gateway calls it. Managed from the Console, never from code.
  • Resolution per request: the agent's default, then the Organization's plan tier, then the user's choice when the plan allows it; a retired profile is replaced by the active one of the same tier.