Skip to content

Hard coded foundational model reference in agent-openai-agents-sdk #188

@omar-khan-manuka-ai

Description

@omar-khan-manuka-ai

Hi there,

When using the Databricks UI to create an agent, if you choose to create a custom agent and deploy the Databricks App, it uses the agent-openai-agents-sdk template to create the agent and the chatbot front-end. Currently the agent is created with this code:

def create_agent(mcp_servers: list[McpServer] | None = None) -> Agent: return Agent( name="Agent", instructions="You are a helpful assistant.", model="databricks-gpt-5-2", tools=[get_current_time], mcp_servers=mcp_servers or [], )

This hard-coded model is not deployed as a foundation model by default and the chatbot fails when you ask a question. Might be worth moving this to the databricks.yml file as a variable and change to an older model?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions