Release Notes

Anthropic and Google models in the AI Agent Action

The AI Agent Action in Workflows now supports Anthropic and Google models alongside OpenAI. The model picker groups models by provider, marks reasoning-capable models with a thinking chip, and gives each model a short description of its intended use.

  • Anthropic: Claude Opus 5, Claude Sonnet 5, Claude Haiku 4.5

  • Google: Gemini 3.6 Flash, Gemini 3.1 Pro Preview

  • OpenAI: GPT-5.6 Luna, GPT-5.6 Tera, GPT-5.6 Sol, GPT-5 Nano

Set reasoning effort to Low, Medium, or High from the same dropdown; the control is available only for models that support thinking. Higher effort gives deeper reasoning for complex tasks, and lower effort gives faster responses for simpler ones. Combined with earlier token optimization, the same task can run up to 50% cheaper.