Tool compatibility and results matrix

Tool compatibility and results matrix defines model validation status interpretations, supported cloud provider requirements, and custom endpoint compliance rules within the Agent Studio model abstraction layer.

Table 1. Interpretation of validation results
Validation status	Interpretation and required actions
`Model supports function calling`	The LLM successfully returned a structured tool call and is fully certified for use in agents and workflows. The UI displays the specific tool invoked, verified arguments, and real-time processing response latency.
`Model does not support function calling`	The LLM failed to return a structured tool call and cannot power agents in Agent Studio. Select an alternative model variant or deployment configuration that natively supports tool calling.
`Test Error`	The request failed due to infrastructure or transport layers, such as an invalid endpoint URL, incorrect credentials, or model provider unavailability. Resolve the underlying connection parameters and retry the test.

Supported model providers and endpoint requirements

Agent Studio communicates with the following major cloud providers and custom infrastructures using a consistent model abstraction layer:

Enterprise cloud providers: OpenAI, Azure OpenAI, Google Gemini, Anthropic, and AWS Bedrock.
During registration for these standard providers, Agent Studio proactively enforces function-calling compatibility and rejects LLMs known to lack these features.
Custom and private deployments: Cloudera AI Inference (CAI) and OpenAI Compatible custom endpoints.

Authoritative validation rule

For Cloudera AI Inference and custom OpenAI-compatible endpoints, function calling, also known as tool calling, cannot always be automatically detected upfront. Use the Function Calling Test to verify compatibility.

If a custom endpoint validation fails, verify the following configuration requirements:

The LLM serving runtime must explicitly enable function-calling or tool-calling modules.
The endpoint must expose a fully OpenAI-compliant chat completions API structure.