Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.gravixlayer.ai/llms.txt

Use this file to discover all available pages before exploring further.

Control and manage every aspect of AI agents while we handle the infrastructure. Code execution, terminal access, filesystem, SSH, and networking — all spun up in ~50 milliseconds. No cold starts. Full hardware-level isolation. Each agent runtime is its own microVM with dedicated CPU, memory, disk, and network. No shared resources. No container reuse. Every agent gets a clean, secure environment — and it runs indefinitely until you terminate it. Install any package, system library, or framework. Fully customize the runtime with custom templates or install packages on the fly with run_cmd.

​
How It Works

from gravixlayer import GravixLayer

client = GravixLayer()

# Create an agent runtime (runtime boots in ~50ms, no cold starts)
runtime = client.runtime.create(template="python-3.14-base-small")

# Execute code
result = runtime.run_code(code="print(2 + 2)")
print(result.text)  # 4

# Run terminal commands
cmd = runtime.run_cmd(command="pip install requests --quiet")
cmd = runtime.run_cmd(command="pip", args=["install", "requests", "-q"])
print(cmd.stdout)

# Clean up
runtime.kill()

​
What You Can Build

AI Agent Code Execution

Execute LLM-generated code in agent runtimes

Tool & MCP Server Hosting

Host MCP servers, APIs, or agent tools

Code Interpreter

Build interpreters that execute and iterate

Data Analysis

Process datasets and generate charts

Reinforcement Learning

Run concurrent runtimes for reward evaluation

AI Evaluations

Run test suites in isolated environments

​
Capabilities

Code Execution

Python and JavaScript with persistent contexts

Terminal

Shell commands, package installs, scripts

File System

Read, write, upload, and download files

Git Operations

Clone, commit, push, and manage branches inside the runtime

Custom Templates

Build environments from any Docker image

Access Runtime

SSH or web terminal access

LLM Integration

Connect LLMs for autonomous execution

​
Templates

Pre-built templates ready to use. Built on python:3.14-slim-bookworm.
TemplatevCPUMemoryDisk
python-3.14-base-small11 GB2 GB
python-3.14-base-medium12 GB4 GB
python-3.14-base-large24 GB8 GB

​
Multi-Cloud

ProviderRegionStatus
azureeastus2 (Virginia)Available
awsus-east-1 (Virginia)Coming Soon
gcpus-east1 (South Carolina)Coming Soon