[yamaceay]

Context

Closes SAP/ai-sdk-python-backlog#76.

What this PR does and why it is needed

This PR proposes a new utility function contributing to a more convenient model discovery using the base model client. The new function wraps the .query(...) method and returns the list of models directly.

Definition of Done

• Code is tested (Unit, Integration, E2E)
• Error handling created / updated & covered by the tests above
• Documentation updated
  • Only Public APIs are allowed to be used in documentation/tutorials/sample code
• (Optional) Aligned changes with the JS/TS and Java SDK
• (Optional) Release notes updated -->

[yamaceay]

Below is a quick demo of the new model discovery helpers. Given that the new base version is released and reused by core package, and we initialize the base client as follows:

from ai_core_sdk.ai_core_v2_client import AICoreV2Client
from ai_api_client_sdk.helpers.llm_helper import (
    filter_foundation_models_by_capability,
    filter_foundation_models_non_deprecated,
    get_foundation_model_latest_versions,
)

client = AICoreV2Client.from_env()

List all available models

models = client.get_models()

Filter by capability

vision_models = filter_foundation_models_by_capability(models, "vision")

Filter out deprecated versions

active_models = filter_foundation_models_non_deprecated(models)

print(f"Models before: {len(models)}")
print(f"Models after removing deprecated: {len(active_models)}\n")

Get latest versions per model

latest = get_foundation_model_latest_versions(models)

print(f"Latest versions for first 5 models:\n")
for entry in latest[:5]:
    version_names = [v.name for v in entry["latest_versions"]]
    print(f"  {entry['model']} ({entry['executable_id']}): {version_names}")

Resolve orchestration deployment

from ai_api_client_sdk.models.status import Status

orch_deployments = client.deployment.query(
    scenario_id="orchestration",
    status=Status.RUNNING,
).resources

Resolve a foundation model deployment

fm_deployments = client.deployment.query(
    scenario_id="foundation-models",
    status=Status.RUNNING,
).resources

if fm_deployments:
    d = fm_deployments[0]
    model_info = (d.details or {}).get("resources", {}).get("backend_details", {}).get("model", {})
    print(f"Deployment ID:  {d.id}")
    print(f"Model:          {model_info.get('name', 'unknown')}")
    print(f"Deployment URL: {d.deployment_url}")
else:
    print("No running foundation model deployment found.")

[OscarDDD]

Overall good and I think this worths an alignment in the team. Just few nitpicks.

[q] In general I also see proxy_client supports some deployment-related functionalities. Will there be any duplicates between ?