Quickstart

Inference Stacks

Create an Adaption API key, install the Python SDK, and generate your first model response.

The Responses API generates model output from text or a list of messages. This guide takes you from a new API key to your first response in a few minutes.

This guide requires Adaption Python SDK version 0.16.0 or later.

Create and export an API key

Sign in to Adaption.
Open Settings, then open API keys.
Create a key and copy it. You will not be able to view the key again.
Export the key as ADAPTION_API_KEY. The SDK reads this variable automatically:
Terminal window
```
export ADAPTION_API_KEY="pt_live_..."
```

Install the SDK

pip install adaption

Send a request

Call responses.create with a model and an input:

from adaption import Adaption

client = Adaption()

response = client.responses.create(
    model="qwen3.5-397b",
    input="Write a one-sentence bedtime story about a unicorn.",
)

print(response.output_text)

output_text joins the text from every message in the response’s output array. The full output array also includes any tool calls the model made.

To guide the model’s behavior, add instructions:

response = client.responses.create(
    model="qwen3.5-397b",
    instructions="You are a helpful assistant.",
    input="Hello!",
)

Next steps

Stream output as it is generated by passing stream=True.
Continue a conversation with store=True and previous_response_id.
Give the model tools, such as MCP servers and file_search over your adapted datasets.

See the responses.create API reference for all request parameters.