Google Agent Platform Code Execution
Google-managed sandbox API (formerly part of Vertex AI Agent Engine) that lets agents deployed anywhere run Python in stateful, network-isolated environments. A tightly locked-down Python tool for Google Cloud users; too constrained for coding agents.
Overview
Google’s managed Code Execution sandbox, now under the Gemini Enterprise Agent Platform (formerly Vertex AI Agent Engine), runs Python in gVisor-isolated containers with no network access. State persists across calls for up to 14 days, which suits conversational analysis tools.
The library set is fixed — you cannot install your own packages — and parts of the product were still labelled preview when we checked.
Notable
- Agents do not have to run on Google Cloud to call it.
- File I/O up to 100 MB per request or response.
Google Agent Platform Code Execution specifications
| Attribute | Details |
|---|---|
| Isolation | gVisor-isolated container in a Google-managed tenant project; does not inherit the calling agent’s credentials |
| Deployment | Cloud provider service |
| Open source | No |
| SDKs and interfaces | Python (Agent Platform SDK), REST API |
| Cold start (vendor claim) | Docs: create and execute code “in under a second”; a cold provision can take up to 2 minutes |
| Max session length | State retained up to 14 days (configurable TTL); each execution times out after 300 s |
| Persistence | Stateful across calls for the TTL; snapshots not supported for Code Execution sandboxes |
| GPU | No |
| Networking | No network access from the code execution sandbox; VPC Service Controls supported |
| Pricing | $0.085 per vCPU-hour + $0.009 per GiB-hour, rounded to the second |
| Free tier | Not documented |
| Compliance | VPC Service Controls, CMEK, data residency at rest and HIPAA per Google’s Agent Platform security table |
| Regions | Not documented |
Facts from Google Agent Platform Code Execution’s public documentation, verified November 8, 2026. Performance numbers are vendor claims.
Strengths and trade-offs
Strengths
- You want a no-network, gVisor-isolated Python tool
- You need state that lasts days, not minutes
- You are on Google Cloud and need CMEK or VPC Service Controls
Trade-offs
- Python only, with a fixed library set
- No network access and no snapshots
- Launch stage partly preview
Best for: Google Cloud agents that need stateful, no-network Python execution.
Where Google Agent Platform Code Execution ranks
How Google Agent Platform Code Execution compares
| Sandbox | Isolation | Open source | Self-host / BYOC | Cold start (vendor) | Max session | Snapshots / pause | GPU | SDKs |
|---|---|---|---|---|---|---|---|---|
| Google Agent Platform Code Execution | gVisor | No | No / No | <1 s (warm) | 14-day state | No / Not documented | No | Python (Agent Platform SDK), REST API |
| Kubernetes Agent Sandbox | gVisor or Kata | Yes | Yes / Yes | <1 s (warm pool) | Not documented | Yes / Yes | Yes | Kubernetes CRDs, Python, kubectl |
| AWS Bedrock AgentCore | MicroVM (Runtime) | No | No / No | Not documented | Up to 8 h | Not documented / Not documented | No | Python (AgentCore SDK), AWS SDKs, AWS CLI |
| Azure Container Apps dynamic sessions | Hyper-V VM | No | No / No | Sub-second (pooled) | 220 s / exec | No / No | No | REST API, Azure CLI, LangChain, LlamaIndex, Semantic Kernel |
| OpenAI Code Interpreter & Hosted Shell | VM (unspecified) | No | No / No | Not documented | 20 min idle | No / No | No | OpenAI Responses API, Containers API, OpenAI SDKs |
Frequently asked questions
What is Google Agent Platform Code Execution?
How does Google Agent Platform Code Execution isolate code?
Is Google Agent Platform Code Execution open source?
How much does Google Agent Platform Code Execution cost?
What are the best Google Agent Platform Code Execution alternatives?
Sources
Cite this page
Agent Sandboxes. (2026). Google Agent Platform Code Execution. agentsandboxes.org. Last reviewed November 8, 2026. https://agentsandboxes.org/sandboxes/google-code-execution/
Found an error or a missing sandbox? Submit a correction — every entry links to the sources it relies on.