Anthropic, Google AI Models Now In AI Agent Action

TL;DR

AI Agent Action in workflows now supports Anthropic and Google models, alongside OpenAI. The model selection interface is updated with new grouping, descriptions, and reasoning effort controls. No immediate action is required unless you plan to integrate these new models into your workflows.

Who this affects

HighLevel has not confirmed who is specifically affected. This update primarily impacts agency owners and SaaS entrepreneurs who utilize the AI Agent Action within HighLevel workflows. If you are leveraging AI agents for customer interactions, lead qualification, or internal process automation, you will find new options available.

What changed

The AI Agent Action in HighLevel workflows has expanded its model support and received a UI update. Specifically:

• Models from Anthropic are now available: Claude Opus 5, Claude Sonnet 5, Claude Haiku 4.5.
• Models from Google are now available: Gemini 3.6 Flash, Gemini 3.1 Pro Preview.
• Models from OpenAI are also listed: GPT-5.6 Luna, GPT-5.6 Tera, GPT-5.6 Sol, GPT-5 Nano.
• The model selection interface has been redesigned for clarity.
• Models are now grouped by their provider (Anthropic, Google, OpenAI) in the interface.
• Models capable of reasoning are now indicated with a 'thinking chip' icon.
• Each model now includes a short description outlining its intended purpose.
• A new control allows setting the 'Reasoning effort' to Low, Medium, or High for models that support this feature.

HighLevel has not confirmed if this update replaces or breaks existing functionalities. Limits or plan gating information has not been confirmed. The rollout status is unknown.

Where to find it

This update is located within the 'AI Agent Action' element inside HighLevel workflows. To access it, navigate to your desired workflow, add or edit an 'AI Agent Action', and then look for the model selection dropdown menu. The redesigned interface and new model options will be present there.

How agencies will use this

Agencies can leverage these new AI models to fine-tune their automated client communications. For instance, an agency managing a client's customer support could create a workflow with an 'AI Agent Action' triggered by incoming support tickets. By selecting Claude Opus 5, known for its strong reasoning capabilities, and setting the 'Reasoning effort' to 'High', the AI agent could analyze complex customer queries, identify sentiment, and draft more nuanced initial responses for human review. Similarly, for lead qualification, an agency might use Gemini 3.6 Flash for quick, cost-effective initial interactions within a workflow, routing more complex conversations to a human agent or a more powerful model based on the initial AI assessment. This allows for more granular control over AI agent performance based on specific task requirements within a sub-account's workflow.

Our take

This is a major update for agencies already using or planning to use AI agents in their HighLevel workflows. The expanded choice of models from Anthropic and Google provides significant flexibility, allowing agencies to select the best tool for a given task, potentially optimizing for cost, speed, or reasoning capability. The UI improvements and reasoning effort control are practical enhancements that make the AI Agent Action more usable and powerful. This is not a minor update; it directly impacts how agencies can build advanced automations.