12.2 Serving Beyond Agent Platform Endpoints: Model Garden, Cloud Run & GKE

Key Takeaways

  • Model Garden serves Google models through managed APIs, partner and open models through MaaS, and open or custom-weight models through one-click self-deployment to dedicated endpoints.
  • Cloud Run services can attach one NVIDIA L4 (24 GB) or RTX PRO 6000 Blackwell (96 GB) GPU per instance, and GPU instances can scale to zero.
  • Cloud Run GPU instances with preinstalled drivers start in about 5 seconds, which suits spiky or low-traffic inference.
  • GKE inference supports model servers such as vLLM, TensorFlow Serving, and Triton, with GKE Inference Gateway, Horizontal Pod Autoscaling, and Inference Quickstart.
  • Choose GKE when you need Kubernetes-level control, portability, or shared accelerator pools, and Agent Platform endpoints when you want managed ML serving features such as Model Registry versions and Model Monitoring.
Last updated: September 2026

The exam guide names Agent Platform, Model Garden, Cloud Run, and GKE as serving options. Picking among them depends on the model source, traffic pattern, operations appetite, and the ML-specific features you need.

Serving Through Model Garden

Model sourceServing pathOperations
Google models (Gemini, Veo, embeddings)Managed API with PayGo, Provisioned Throughput, or batchNone
Partner and open models via MaaSServerless API ("API Service" model cards)None
Open models with your weights or tuningOne-click self-deployment to a dedicated endpoint in your project, using prebuilt serving containers (vLLM, Hex-LLM, SGLang, TGI, TensorRT-LLM) or a custom vLLM containerYou choose machine type and accelerators and pay for compute while deployed

Self-deployed endpoints support Agent Platform features such as dedicated or private networking and the Chat Completions API for Model Garden models. Model Garden also offers batch inference for supported models. Before deploying, use the SDK's list_deploy_options() to see verified machine and accelerator configurations.

Serving on Cloud Run

Cloud Run runs stateless containers with request-based autoscaling, and it can scale to zero.

CapabilityDetail
GPUsNVIDIA L4 (24 GB) or RTX PRO 6000 Blackwell (96 GB). One GPU per instance
Minimum resourcesL4: at least 4 CPU and 16 GiB memory. RTX PRO 6000: at least 20 CPU and 80 GiB
StartupInstances with GPU drivers preinstalled start in about 5 seconds
ScalingOn-demand GPUs with no reservation needed, and scale to zero when idle
RegionsLimited to specific regions for each GPU type

Cloud Run fits well when:

  • Traffic is spiky or low, and idle cost should be near zero.
  • The model fits on one GPU or runs on CPU (small LLMs, embedding models, scikit-learn, XGBoost).
  • The team already deploys containerized web services and wants the same workflow for models.
  • You need a custom HTTP API around the model (authentication, business logic).

Trade-offs: you build and maintain the serving container, and cold starts include loading model weights (so keep images and weights small, or load weights from Cloud Storage efficiently). Cloud Run can split traffic between revisions, but ML-specific features such as built-in Model Monitoring, Model Registry version deployment, and explanations aren't provided the way Agent Platform endpoints provide them. You manage model versioning yourself.

Serving on GKE

GKE runs model servers as Kubernetes Deployments.

ComponentRole
Model serversvLLM, JetStream, TensorFlow Serving, Triton Inference Server, or a custom Python server
GKE Inference GatewayLoad balancing and routing tailored to gen AI inference
Horizontal Pod Autoscaler (HPA)Scale pods on CPU, latency, or custom metrics
GKE Inference QuickstartRecommended accelerators, model servers, scaling, and storage configurations based on performance and cost goals
Autopilot vs. StandardAutopilot manages nodes, including accelerators. Standard gives full node-pool control
EcosystemKueue for queueing, Ray for distributed compute, LeaderWorkerSet for multi-host serving, LoRA adapters

GKE supports models from about 8 billion to 671 billion parameters on GPUs and TPUs, with dashboards for popular servers such as vLLM.

GKE fits well when:

  • The organization runs Kubernetes already and wants one platform for apps and models.
  • Portability across environments matters.
  • You need fine-grained control: custom schedulers, multi-host serving of very large models, or sharing accelerator pools across many models.
  • Very high, steady volume makes hand-tuned serving cost-effective.

Trade-offs: you operate clusters, upgrades, security, and scaling policies.

Choosing a Serving Platform

RequirementBest fit
Managed ML serving with model versions, traffic splits, Model Monitoring, and explanationsAgent Platform endpoints
Use Gemini or a partner/open model with no infrastructureModel Garden API or MaaS
Serve your own weights for an open LLM with managed endpoint featuresModel Garden self-deployment
Spiky traffic, scale to zero, single-GPU or CPU models, container-native teamCloud Run
Kubernetes-standardized org, portability, shared GPU/TPU pools, multi-host giant modelsGKE
Batch scoring of data in BigQueryBigQuery ML or Agent Platform batch inference

Operational Checklist for Any Serving Platform

ConcernAgent Platform endpointCloud RunGKE
Versioned rolloutTraffic split across deployed modelsRevision traffic splitDeployment strategies or Gateway routing
AutoscalingMin/max replicas with CPU, GPU, and request metricsRequest concurrency, scale to zeroHPA on custom metrics
Private accessPrivate Service Connect or private endpointsInternal ingress, VPC connectorsInternal load balancers
Prediction logging for monitoringBuilt-in request-response loggingBuild it yourselfBuild it yourself
Model MonitoringBuilt inCustomCustom

Worked Scenarios

  1. An internal document classifier gets 200 requests a day during business hours, with a small fine-tuned transformer on one L4. Cloud Run with an L4 GPU scales to zero overnight, so idle cost is minimal.
  2. A bank serving a fraud model needs canary rollouts between versions, drift monitoring, and private networking. Agent Platform endpoint (Private Service Connect) with traffic splitting and Model Monitoring.
  3. A platform team serves 40 open LLMs and adapters to many product teams on shared H100 pools with custom routing. GKE with vLLM, Inference Gateway, and HPA.
  4. A startup wants a Llama chat model with no GPU management. MaaS from Model Garden.
Test Your Knowledge

An internal tool calls a small fine-tuned transformer about 150 times a day, only during business hours, and leadership wants near-zero cost when idle. The model fits on one L4 GPU. Which serving option fits best?

A
B
C
D
Test Your Knowledge

A platform team must serve dozens of open LLMs and LoRA adapters for many product teams on shared GPU pools, with portable Kubernetes manifests and custom routing. Which platform fits best?

A
B
C
D
Test Your Knowledge

A company wants to use a partner LLM from Model Garden for a new feature with no GPU provisioning and pay-per-use billing. Which serving path should it choose?

A
B
C
D