Using Downsizing with PydanticAI

Prerequisites

Before configuring PydanticAI, make sure your Downsizing account can make requests — pay-per-use or a connected subscription both work. PydanticAI ships its model integrations behind optional extras on the pydantic-ai-slim package; install the one for the API you want to use:

pip install "pydantic-ai-slim[anthropic]"

pip install "pydantic-ai-slim[openai]"

(Installing the full pydantic-ai package pulls in both, plus every other provider extra.)

Option A: Anthropic-compatible endpoint (recommended)

PydanticAI's AnthropicProvider wraps an anthropic SDK client, so point that client's base_urlat Downsizing's Anthropic-compatible endpoint and pass it into the provider:

from anthropic import AsyncAnthropic
from pydantic_ai import Agent
from pydantic_ai.models.anthropic import AnthropicModel
from pydantic_ai.providers.anthropic import AnthropicProvider

client = AsyncAnthropic(
    api_key="dws_YOUR_API_KEY",
    base_url="http://localhost:3001/YOUR_INFERENCE_ID/anthropic",
)

model = AnthropicModel(
    "claude-opus-4-8",
    provider=AnthropicProvider(anthropic_client=client),
)

agent = Agent(model)

Pass the resulting model to any Agent in your application, or override it per-run via agent.run(..., model=model).

Option B: OpenAI-compatible endpoint

For OpenAI-style models, use OpenAIResponsesModel, which targets the Responses API, together with OpenAIProvider. Set base_urlto Downsizing's OpenAI-compatible endpoint:

from pydantic_ai import Agent
from pydantic_ai.models.openai import OpenAIResponsesModel
from pydantic_ai.providers.openai import OpenAIProvider

model = OpenAIResponsesModel(
    "gpt-5",
    provider=OpenAIProvider(
        base_url="http://localhost:3001/YOUR_INFERENCE_ID/openai/v1",
        api_key="dws_YOUR_API_KEY",
    ),
)

agent = Agent(model)

PydanticAI also ships OpenAIChatModel, which targets the older Chat Completions API — prefer OpenAIResponsesModelabove when using Downsizing's OpenAI-compatible endpoint.

Configuration

Find your Inference ID and API key on the Inference API page. Replace YOUR_INFERENCE_ID with your workspace UUID and dws_YOUR_API_KEY with your Downsizing API key in either snippet above.

Verify

Run the agent to confirm traffic is routing through Downsizing:

result = agent.run_sync("Say hello in one short sentence.")
print(result.output)

If you get a response back, Downsizing is set up correctly. You can also check your savings dashboard to see the request appear in your usage metrics.