[Inference] Gemini connector error due to long running request - timeout of 60000ms exceeded

[qn895]

Kibana version:

Gemini connector's invokeStream API occasionally has such long running request that it times out:

[2025-06-16T15:44:54.611-04:00][WARN ][plugins.actions.gemini] action execution failure: .gemini:gemini-25-pro: gemini-2.5-pro-preconfigured: an error occurred while running the action: Status code: undefined. Message: Unexpected API Error: ECONNABORTED - timeout of 60000ms exceeded; retry: true
[2025-06-16T15:45:58.656-04:00][ERROR][plugins.actions.gemini] Error: Status code: undefined. Message: Unexpected API Error: ECONNABORTED - timeout of 60000ms exceeded
    at GeminiConnector.getRequestError (sub_action_connector.ts:193:29)
    at GeminiConnector.request (sub_action_connector.ts:180:26)
    at GeminiConnector.streamAPI (gemini.ts:262:22)
    at GeminiConnector.invokeStream (gemini.ts:374:13)
    at Object.executor (executor.ts:99:18)
    at action_executor.ts:504:23
    at ActionExecutor.execute (action_executor.ts:153:12)
    at Object.invoke (inference_executor.ts:44:14)

To reproduce:

1. Run

node x-pack/solutions/observability/plugins/observability_ai_assistant_app/scripts/evaluation/index.js --persist --clear --evaluateWith "other"

2. Pick any pre-configured connector
3. Pick gemini 2.5 pro for the evaluateWith model

We need to investigate what's causing the endpoint to take so long.

Elasticsearch version:

Server OS version:

Browser version:

Browser OS version:

Original install method (e.g. download page, yum, from source, etc.):

Describe the bug:

Steps to reproduce:
1.
2.
3.

Expected behavior:

Screenshots (if relevant):

Errors in browser console (if relevant):

Provide logs and/or server output (if relevant):

Any additional context:

[elasticmachine]

Pinging @elastic/ml-ui (:ml)