Vertex AI Agent Builder, rebranded as the Gemini Enterprise Agent Platform at Google Cloud Next 2026, is a pay-as-you-go platform for building and deploying enterprise AI agents. It bundles a code-first ADK, a low-code visual Agent Studio, 200+ foundation models (Gemini, Claude on Vertex, Llama, and others), a managed runtime (Agent Engine), persistent memory, and enterprise governance. Pricing is composed across four separate meters: runtime compute, session events, search queries, and per-token model costs. Existing Vertex AI customers do not need to migrate. Key features: - Agent Development Kit (ADK) for code-first agent development - Low-code visual Agent Studio with drag-and-drop interface - 200+ foundation models including Gemini 3 Pro, Claude, and Llama via Model Garden - Agent Engine managed runtime with persistent memory and session management - Retrieval Augmented Generation (RAG) with Vertex AI Search - Integration with LangChain and LlamaIndex frameworks
Pay-as-you-go: Agent Engine runtime at $0.0864/vCPU-hour; memory at $0.009/GB-hour; session/Memory Bank events at $0.25/1,000; Vertex AI Search from $1.50 to $6.00/1,000 queries. Model tokens billed separately per model. $300 in free credits for new Google Cloud customers.