Model-Aware Reasoning Effort for Voice Agents
*Category:* Enhancement
*Location:* Voice Agents → Create or Edit Voice Agent → Advanced Settings → LLM Configuration
*Release Date:* September 1 2026
What’s new
Voice Agents now show a *Reasoning Effort* selector when the chosen LLM supports it.
Available options come from that specific model, so unsupported effort levels are never offered.
Switching models automatically preserves a valid saved effort or selects the lowest active supported effort when necessary.
Models without configurable reasoning hide the control.
How to use it
*Advanced Settings* must be enabled on the account. If it is not visible, Customer Support can enable it or make the change for you.
Open a Voice Agent, go to *Advanced Settings → LLM Configuration*, and choose a supported LLM.
Set *Reasoning Effort* from the options shown for that model, then save.
Impact
Teams can tune reasoning depth for supported LLMs while keeping Voice Agent configuration valid as model capabilities change.