Vertex AI is Google Cloud’s managed platform for building, training, and serving ML/AI models — and, as of Google Cloud Next 2026, its identity has shifted decisively toward agents. Google rebranded it to the Gemini Enterprise Agent Platform, consolidating it with Agentspace into a single product. Existing Vertex AI customers don’t need to migrate anything — the platform and APIs continue to work, the umbrella name changed. This note stays focused on the deploy runtime and pricing; the full build/scale/govern/optimize lifecycle and the new governance stack (Agent Identity, Agent Registry, Agent Gateway) live in What is Gemini Enterprise Agent Platform.
What actually changed
Before 2026: Vertex AI (ML platform) + Agentspace (separate agent product)
2026: Gemini Enterprise Agent Platform (one product)
├── Agent Builder — build agents
└── Deployments — run agents (formerly "Agent Engine")
Agent Builder: two ways to build
| Mode | Who it’s for | How it works |
|---|---|---|
| Agent Studio | Low-code / business users | Visual builder over Google’s foundation models |
| What is Google ADK | Engineers who want code-first control | Open-source framework, deployed onto this platform |
Both paths bundle access to 200+ foundation models, a managed runtime, and governance controls (tool allow-listing, audit, access policy) in one pay-as-you-go service — the governance layer is the part most enterprise adopters actually pay for, not the models themselves.
Deployments (formerly Agent Engine)
This is the managed runtime layer — where an agent actually lives and takes requests, as opposed to where it was authored.
Your ADK / LangGraph / custom agent code
│
▼
Deployments (managed runtime)
│
┌──────┴──────┐
▼ ▼
Session state Memory Bank
(short-term) (long-term, GA)
Memory Bank reaching General Availability in 2026 means this runtime now manages both short-term session state and long-term memory for production workloads directly — functionally overlapping with what What is Mem0 does as a standalone layer, except scoped to agents deployed on this specific runtime.
Pricing shape (2026)
Worth knowing before recommending this as a deploy target, since the billing model is metered compute + memory, not per-request:
| Meter | Rate |
|---|---|
| Agent Engine runtime (vCPU) | $0.0864 / vCPU-hour |
| Agent Engine runtime (memory) | $0.0090 / GB-hour |
| Free tier | 50 vCPU-hours + 100 GB-hours/month |
| Stored session events / memories (from Jan 28, 2026) | $0.25 per 1,000 events/memories |
The practical implication: an idle-but-deployed agent still bills for the runtime keeping it alive, the same cost model as a running container rather than a serverless function that scales to zero.
Where it fits
| Layer | Component |
|---|---|
| Build (code-first) | What is Google ADK |
| Build (low-code) | Agent Studio |
| Run | Deployments (managed runtime + Memory Bank) |
| Data access for agents | What is MCP Toolbox (databases), What is Grafana MCP (observability data) |
Why it’s on the backlog: it’s the “where does this actually run in production” answer for anything built with What is Google ADK — worth evaluating against self-hosting What is Hermes Agent/What is CrewAI on your own infra when the governance and managed-memory features outweigh the metered-compute cost.
Local graph
Linked from 7 notes
What is Azure AI Services
Microsoft's Azure AI service catalog — account models (single- vs multi-service), Azure OpenAI's deployment-based access pattern, Azure AI Search as the RAG grounding layer, and the single-service capability catalog (Vision, Language, Speech, Document Intelligence).
What is Mem0
Universal memory layer for AI agents — combines vector search, a knowledge graph, and key-value caching behind one API, so any framework can bolt on persistent, cross-session memory in under a day.
What is CrewAI
Open-source Python framework for multi-agent orchestration — role-based 'Crews' for autonomous collaboration and event-driven 'Flows' for precise control, now a de facto standard for production agentic pipelines.
What is Gemini Enterprise Agent Platform
Google's April 2026 unification of agent tooling — a four-stage lifecycle (build, scale, govern, optimize) wrapping Agent Studio/ADK, a stateful Agent Runtime, an Identity/Registry/Gateway governance stack, and native A2A + MCP interop.
What is Google ADK
Google's open-source, code-first Agent Development Kit — a multi-language framework for building, evaluating, and deploying agents, positioned as an 'agent execution framework' rather than a toolkit.
Agentic AI: Projects & Engineering Mastery
A book-shaped table of contents for Agentic AI: Projects & Engineering Mastery: hands-on practitioner builds, Principal/Staff-level technical leadership, and the lookup appendices and vendor/framework reference notes for the whole series. Book 6 of the AI Systems Engineering series.
7. Google ADK
Covers Google's Agent Development Kit — its workflow/dynamic-routing composition model, native tool-integration story, and the deployment path onto Vertex AI Agent Engine that is the actual site of vendor coupling, not the framework code itself.
Related notes
What is Azure AI Services
Microsoft's Azure AI service catalog — account models (single- vs multi-service), Azure OpenAI's deployment-based access pattern, Azure AI Search as the RAG grounding layer, and the single-service capability catalog (Vision, Language, Speech, Document Intelligence).
What is Gemini Enterprise Agent Platform
Google's April 2026 unification of agent tooling — a four-stage lifecycle (build, scale, govern, optimize) wrapping Agent Studio/ADK, a stateful Agent Runtime, an Identity/Registry/Gateway governance stack, and native A2A + MCP interop.
What is Google ADK
Google's open-source, code-first Agent Development Kit — a multi-language framework for building, evaluating, and deploying agents, positioned as an 'agent execution framework' rather than a toolkit.
What is MCP Toolbox
Google's open-source MCP server for databases (formerly Gen AI Toolbox for Databases) — production-ready (v1.0.0) prebuilt tools that connect agents and IDEs directly to AlloyDB, Spanner, Cloud SQL, BigQuery, and more, with OAuth2 zero-trust gating.