[lubaihua33]

Update the AI model to GPT-5.4 in AI log agent. The accuracy is 93% using GPT5.4

[Copilot]

🟡 Changes recommended

The runbook still documents GPT-4 prerequisites and old deployment defaults, so users who omit optional deployment settings may encounter analysis failures because the requested gpt-5.4 deployment is unavailable.

1 open issue

• Document the required gpt-5.4 deployment default · New

What changed in this PR

Updates LISA’s AI log-analysis defaults to use the Azure deployment named gpt-5.4 for both general and software analysis.

Changes:

• Updates both deployment defaults in the log-agent notifier configuration.
• Aligns the analyze() function defaults with the notifier while preserving explicit overrides.

Suggested validation (not run): Enable the notifier and exercise a controlled failed-test notification with defaults and explicit overrides. A passing smoke test alone will not exercise AI analysis.

Key Test Cases:
smoke_test

Impacted LISA Features:
None — changes are limited to AI analysis and notifier defaults.

Tested Azure Marketplace Images:

• canonical 0001-com-ubuntu-server-jammy 22_04-lts-gen2 latest

File	Description
lisa/​notifiers/​log_agent.py	Sets both notifier deployment defaults to gpt-5.4.
lisa/​ai/​log_agent.py	Sets both analyze() deployment defaults to gpt-5.4.

_{Review effort: Balanced}