Secure sandboxes for multi-tenant workloads

Northflank Sandboxes

Run untrusted code at scale with microVMs. In our cloud or your VPC.

Run 10,000+ isolated workloads

Agents write code. You need somewhere safe to run it.

Run Northflank sandboxes on our infrastructure or inside your own VPC. We handle the hard parts: orchestrating microVMs, managing tenants, scaling workloads, all via API.

Sub-second cold starts

Boot a microVM in under a second. Spin up isolated environments instantly for code execution, agent tasks, or parallel jobs.

True multi-tenant isolation

Every workload runs in its own microVM with Kata Containers or gVisor. In your VPC.

Scale to millions

We’ve been running millions of microVMs monthly since 2021. Horizontal autoscaling built-in. Automatic load balancing and intelligent bin-packing for maximum density.

More than just sandboxes

Full infrastructure for production AI platforms. Package management, persistent storage, GPU support, and built-in CI/CD. Everything needed to run agents, databases, services, inference, training, and more.

Lightning-fast isolation for AI workloads

Built for the agent era

AI agents write code. Code interpreters run untrusted input. Your users want to execute arbitrary scripts. You need isolation that’s actually isolated, with the performance to match.

Execute LLM-generated code safely

VM-level isolation keeps malicious code away from host systems and other tenants. Built for coding assistants, code generation tools, and anything that runs code you didn’t write.

Parallel workers & background jobs

Spawn hundreds of isolated workers simultaneously for parallel processing. Run background jobs, scheduled tasks, and long-running agents with automatic scaling. Resources provision only while workloads run.

Remote agent execution at scale

Each agent gets its own sandbox with dedicated resources. Works at prototype scale. Works at thousands of concurrent sessions.

Persistence options for every workload

Ephemeral or stateful, you choose

Sandboxes can be ephemeral or persistent. Attach volumes for state that survives restarts, mount S3 buckets for artifacts, or deploy stateful databases alongside your sandboxes. Multi-read-write volumes for distributed workloads.

Persistent volumes

Attach fast volumes to sandboxes for state that survives restarts. Start at 4GB, scale to 64TB. Multi-read-write support for distributed workloads sharing state.

S3-compatible object storage

Attach S3 buckets or deploy MinIO for code artifacts, model outputs, and generated files. Seamless integration with sandbox workloads for reading inputs and writing results.

Stateful databases

Deploy Redis, Postgres, MySQL, or MongoDB alongside your sandboxes. Perfect for storing agent memory, execution history, and shared state across sandbox instances.

Ephemeral by default

Run sandboxes ephemerally for stateless workloads, or make them persistent when you need state. No forced time limits—run for seconds or weeks depending on your use case.

Testimonials from Sentry

From the front lines

Northflank is way easier than gluing a bunch of tools together to spin up apps and databases. It’s the ideal platform to deploy containers in our cloud account, avoiding the brain damage of big cloud and Kubernetes. It’s more powerful and flexible than traditional PaaS – all within our VPC. Northflank has become a go-to way to deploy workloads at Sentry.

David Cramer

Co-Founder and CPO @ Sentry

Read a testimonial from Sentry

Everything where you expect it to be

Optimized for developer happiness

We designed every part of Northflank with one goal: simplify complex infrastructure and give developers the controls they need. So they can focus on what they do best: building.

Built for scale

You’re in good company

Since 2019, teams have used Northflank to run everything from enterprise products to high-scale AI infrastructure. Whether it’s one container or one thousand, Northflank holds the line.

Millions

Containers

130B+

Requests processed

$24M+

Raised in funding

50k+

Developers in production

330+

Availability zones

Deploy anywhere, control everything

Flexible deployment

Bring-your-own-cloud deployment in your VPC or use our managed infrastructure. Multi-region support with consistent APIs. GPU acceleration for compute-intensive AI workloads. Built-in observability, CI/CD, and cost tracking.

Horizontal autoscaling

Set minimum and maximum sandbox counts, let autoscaling handle demand spikes. Scale based on CPU, memory, and RPS thresholds. Bin-packing maximizes density without breaking isolation.

GPU support

H100s and other GPUs for inference, training, and compute-intensive agent work. $2.74/hour all-in, up to 62% cheaper than the big clouds. Full GPU isolation inside microVMs.

Cost controls

Set spending alerts and monitor expenses at the workload level. CPU at $0.01667/vCPU-hour, up to 65% cheaper than competitors. Track sandbox costs individually or aggregate across your entire platform.

Built-in CI/CD

Deploy updates automatically. Run tests in isolated environments. Version control integration, automatic rollbacks.

Testimonials from Pebblebed