tiered_model_factory.py
"""Tiered Model Factory -- model selection based on subscription tier.
Demonstrates per-tenant model selection: enterprise tenants get the best model,
free-tier users get a cheaper one. The tier comes from JWT claims (trusted),
so clients can't self-upgrade by changing a request field.
Run:
.venvs/demo/bin/python cookbook/05_agent_os/factories/agent/04_tiered_model_factory.py
Test:
# Free tier (cheaper model)
curl -X POST http://localhost:7777/agents/tiered-agent/runs \
-H "Authorization: Bearer <FREE_TOKEN>" \
-F 'message=Explain quantum computing in one sentence' \
-F 'stream=false'
# Enterprise tier (best model)
curl -X POST http://localhost:7777/agents/tiered-agent/runs \
-H "Authorization: Bearer <ENTERPRISE_TOKEN>" \
-F 'message=Explain quantum computing in one sentence' \
-F 'stream=false'
"""
from datetime import UTC, datetime, timedelta
import jwt as pyjwt
from agno.agent import Agent, AgentFactory
from agno.db.postgres import PostgresDb
from agno.factory import RequestContext
from agno.models.openai import OpenAIResponses
from agno.os import AgentOS
from agno.os.middleware import JWTMiddleware
# ---------------------------------------------------------------------------
# Config
# ---------------------------------------------------------------------------
JWT_SECRET = "a-string-secret-at-least-256-bits-long"
db = PostgresDb(
id="tiered-model-db",
db_url="postgresql+psycopg://ai:ai@localhost:5532/ai",
)
TIER_MODELS = {
"free": "gpt-4.1-mini",
"pro": "gpt-4.1",
"enterprise": "gpt-5.4",
}
TIER_INSTRUCTIONS = {
"free": "You are a helpful assistant. Keep responses brief (2-3 sentences max).",
"pro": "You are a helpful assistant. Provide detailed, well-structured responses.",
"enterprise": (
"You are a premium assistant. Provide comprehensive, expert-level responses. "
"Use examples, cite reasoning, and anticipate follow-up questions."
),
}
# ---------------------------------------------------------------------------
# Factory
# ---------------------------------------------------------------------------
def build_tiered_agent(ctx: RequestContext) -> Agent:
"""Build an agent with model quality based on the caller's subscription tier."""
claims = ctx.trusted.claims
tier = claims.get("tier", "free")
# Fall back to free tier for unknown values
model_id = TIER_MODELS.get(tier, TIER_MODELS["free"])
instructions = TIER_INSTRUCTIONS.get(tier, TIER_INSTRUCTIONS["free"])
return Agent(
model=OpenAIResponses(id=model_id),
instructions=instructions,
add_datetime_to_context=True,
markdown=True,
)
tiered_factory = AgentFactory(
db=db,
id="tiered-agent",
name="Tiered Assistant",
description="Model quality scales with subscription tier (free/pro/enterprise)",
factory=build_tiered_agent,
)
# ---------------------------------------------------------------------------
# AgentOS
# ---------------------------------------------------------------------------
agent_os = AgentOS(
id="factory-tiered-demo",
description="Demo: subscription-tier-based model selection",
agents=[tiered_factory],
)
app = agent_os.get_app()
# Standard JWTMiddleware -- request.state.claims is set automatically
app.add_middleware(
JWTMiddleware,
verification_keys=[JWT_SECRET],
algorithm="HS256",
user_id_claim="sub",
validate=False, # Set True in production
)
# ---------------------------------------------------------------------------
# Run
# ---------------------------------------------------------------------------
if __name__ == "__main__":
def make_token(tier: str, org_id: str = "acme", user_id: str = "user_1") -> str:
payload = {
"sub": user_id,
"tier": tier,
"org_id": org_id,
"exp": datetime.now(UTC) + timedelta(hours=24),
"iat": datetime.now(UTC),
}
return pyjwt.encode(payload, JWT_SECRET, algorithm="HS256")
print("Test tokens (valid for 24h):")
print()
print(f" FREE: {make_token('free')}")
print(f" PRO: {make_token('pro')}")
print(f" ENTERPRISE: {make_token('enterprise')}")
print()
agent_os.serve(app="04_tiered_model_factory:app", port=7777, reload=True)
Run the Example
Set up your virtual environment
uv venv --python 3.12
source .venv/bin/activate
uv venv --python 3.12
.venv\Scripts\activate
Export your API keys
export JWT_VERIFICATION_KEY="your_jwt_verification_key_here"
export OPENAI_API_KEY="your_openai_api_key_here"
$Env:JWT_VERIFICATION_KEY="your_jwt_verification_key_here"
$Env:OPENAI_API_KEY="your_openai_api_key_here"
Run PgVector
docker run -d \
-e POSTGRES_DB=ai \
-e POSTGRES_USER=ai \
-e POSTGRES_PASSWORD=ai \
-e PGDATA=/var/lib/postgresql/data/pgdata \
-v pgvolume:/var/lib/postgresql/data \
-p 5532:5432 \
--name pgvector \
agnohq/pgvector:18