Skip to main content

OpenAI Extension for the TRAC Model Runtime

This extension makes the OpenAI Python SDK available to use from inside a TRAC model.

  • Use the native OpenAI client classes directly in TRAC model code
  • Connection settings managed by the runtime for both local and deployed models
  • Supports both OpenAI and AzureOpenAI clients

Models that make external calls are not considered repeatable, and will be flagged as not repeatable when they run on the TRAC platform.

Installing

The OpenAI extension can be installed with pip:

$ pip install tracdap-ext-openai

The package has the following dependencies:

  • tracdap-runtime (version 0.10.0-beta2 or later)
  • openai (version 1.x)

Using the OpenAI client

Here is a minimum working example of a TRAC model using the OpenAI client:

import tracdap.rt.api as trac
import openai

class OpenAIModel(trac.TracModel):

    # ... define parameters, inputs and outputs

    def define_resources(self):

        return {
            "openai": trac.define_external_system("openai", openai.OpenAI),
        }

    def run_model(self, ctx: trac.TracContext):

        with ctx.get_external_system("openai", openai.OpenAI) as client:

            response = client.responses.create(
                model="gpt-4o",
                instructions="You are a coding assistant that talks like a pirate.",
                input="How do I check if a Python object is an instance of a class?",
            )

            ctx.log.info(response.output_text)

if __name__ == '__main__':
    import tracdap.rt.launch as launch
    launch.launch_model(OpenAIModel, "config/job_config.yaml", "config/sys_config.yaml")

To make this example work, you will need to add openai as a resource in the system config file:

resources:

  openai:
    resourceType: EXTERNAL_SYSTEM
    protocol: openai

The client can be customized by setting additional properties on the resource, which are passed through to the OpenAI client.

resources:

  openai:
    resourceType: EXTERNAL_SYSTEM
    protocol: openai
    properties:
      project: proj_xxxxxxxxxxxxx

The following configuration properties are supported:

  • api_key, string, required
  • organization, string, optional
  • project, string, optional
  • base_url, string, default = https://api.openai.com/v1/
  • timeout, float, defeault = openai.DEFAULT_TIMEOUT.read (currently 600 seconds)
  • max_retries, int, default = openai.DEFAULT_MAX_RETRIES (currently 2)

The api_key should not be put into a config file in plain text, for local development it is recommended to set the OPENAI_API_KEY environment variable instead. If both the config property and the environment variable are set, the config property takes precedence.

Using the AzureOpenAI client

Here is a minimum working example of a TRAC model using the AzureOpenAI client. This assumes the required resources and deployments have been set up in Azure.

import tracdap.rt.api as trac
import openai

class TestModel(trac.TracModel):

    # ... define parameters, inputs and outputs

    def define_resources(self):

        return {
            "openai_azure": trac.define_external_system("openai", openai.AzureOpenAI)
        }

    def run_model(self, ctx: trac.TracContext):

        with ctx.get_external_system("openai_azure", openai.AzureOpenAI) as client:

            completion = client.chat.completions.create(
                model="gpt-4.1-mini",
                messages=[
                    { "role": "system", "content": "You are a coding assistant that talks like a pirate."},
                    { "role": "user", "content": "How do I check if a Python object is an instance of a class?" },
                ]
            )

            ctx.log.info(completion.choices[0].message.content)

if __name__ == '__main__':
    import tracdap.rt.launch as launch
    launch.launch_model(TestModel, "config/job_config.yaml", "config/sys_config.yaml")

To make this example work, you will need to add openai_azure as a resource in the system config file:

resources:

  openai_azure:
    resourceType: EXTERNAL_SYSTEM
    protocol: openai
    subProtocol: azure
    properties:
      api_version: 2025-04-01-preview
      azure_endpoint: https://my-azure-endpoint.cognitiveservices.azure.com/

Setting supProtcol: azure is required for to create an Azure client. The api_version and azure_endpoint properties must be specified, and model parameter in the client call must refer to live model deployment on that endpoint.

All the configuration properties supported by the regular client are also supported by the Azure client. Additionally, the Azure client supports these extra properties:

  • api_version, string, required
  • azure_endpoint, string, required
  • azure_deployment, string, optional
  • azure_ad_token, string, optional

For he Azure client, if api_key is not specified in the config file it is read from the environment variable AZURE_OPENAI_API_KEY. Similarly, azure_ad_token can be read from the environment variable AZURE_OPENAI_AD_TOKEN. If both the config property and the environment variable are set, the config property takes precedence.

Metadata

Release files for tracdap-ext-openai 0.11.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for tracdap-ext-openai 0.11.0
File Size Uploaded Provenance Actions
tracdap_ext_openai-0.11.0.tar.gz 13.2 kB With provenance Details for tracdap_ext_openai-0.11.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for tracdap-ext-openai 0.11.0
File Interpreter ABI Platform Actions
tracdap_ext_openai-0.11.0-py3-none-any.whl Python 3 none any Details for tracdap_ext_openai-0.11.0-py3-none-any.whl

Total release size: 24.8 kB

Release files / tracdap_ext_openai-0.11.0.tar.gz

Download URL tracdap_ext_openai-0.11.0.tar.gz
Size 13.2 kB
Tags Source
SHA-256 checksum
How to use checksums
a99f8046125c436b4075e2ba5528bad066fb783817628a615f8dbad06dbef0ce
BLAKE2b-256 checksum
How to use checksums
4cea72898cf910f9b4d86f006b428cc7d5c531909e93865e0c994f743a865b60
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 19, 2026.

Transparency log

Release files / tracdap_ext_openai-0.11.0-py3-none-any.whl

Download URL tracdap_ext_openai-0.11.0-py3-none-any.whl
Size 11.6 kB
Tags Python 3
SHA-256 checksum
How to use checksums
f6e46eb5306993aabe02181eaf2d073434220827fcadb4aff43822e8bcd85cd1
BLAKE2b-256 checksum
How to use checksums
986890c9aae048ba11f5d4433c6e9310332d5428c1d7856ad4b7e7282de8aecb
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 19, 2026.

Transparency log

Security for tracdap-ext-openai 0.11.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

Both files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for tracdap_ext_openai-0.11.0.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 19, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page