Improved response times and token usage

Get faster responses at lower token cost, automatically. The agent routes specific tasks to faster, specialized models and removes extra LLM calls that inflate token usage, benefiting every Agent Builder user by default.

**What the feature is**
The agent optimizes model selection for specific tasks to enable faster responses at lower token cost. Architectural changes reduce and remove extra LLM calls that expand token usage, and route specific tasks like query generation to faster models with more specialized context.

**Who**
Who: Every Agent Builder user, with the improvements applied by default and no configuration required.

**Why it matters**
Responses get faster and token costs drop at the same time, for every user, without anyone having to opt in or tune anything.

**Key user stories / use cases**
high-volume agents, cost-sensitive deployments, latency-sensitive workflows.