Northflank review

Northflank is a managed runtime platform with a control plane for building and operating containerized workloads. It runs on Northflank Cloud or in customer environments, including cloud accounts and VPCs, on-premises infrastructure and bare metal. It is not an application library: teams connect repositories or container images, then use the platform to deploy and manage services, jobs, databases, sandboxes and GPU workloads.

The platform sits between application code and the underlying compute. Its control plane covers build and release workflows, autoscaling, networking, SSO, RBAC, secrets, logs, metrics, alerts, backups and recovery. Northflank describes workloads as running in isolated microVMs, including agent tasks and untrusted code. That makes it relevant to teams shipping ordinary APIs and workers as well as products that execute user code or run coding agents. For BYOC, workloads run in the customer’s infrastructure while Northflank supplies the managed platform experience.

Northflank’s homepage links to API docs and shows JavaScript examples using an apiClient to create services and execute commands in a sandbox. The examples cover creating a deployment from an image and deploying a service from a GitHub repository with build settings and instances. The pricing page also includes the Platform SDK with Pay as you go. The vendor does not name the API protocol or list other SDK languages on its homepage; confirm SDK support in the vendor’s docs before choosing a language-specific integration.

The deployment surface includes Git repositories and container images, with Dockerfile or buildpacks listed for workloads. The homepage names Codex, Claude Code, OpenCode and Pi as agent harnesses. It also describes previews, pipelines, rollbacks, templating and managed databases. No CLI is named on the homepage, so teams planning command-line workflows should confirm the supported tooling in the docs. The integration model is platform-centric: workloads, releases and infrastructure are managed through Northflank’s control plane rather than a standalone runtime library.

Sandbox is free and includes two services, two jobs, one database, up to $20/month of Northflank Cloud resources and one BYOC cluster capped at 12 vCPU and 48 GB. It also includes unlimited users and no sleeping. That is useful for small workloads, but the included cloud-resource allowance is the boundary to watch; additional usage is billed separately.

Pay as you go starts at $0.01667 per vCPU hour, billed to the second, with no seat fees. Northflank Cloud memory is $0.00833 per GB hour. Compute plans start at $12/month for 0.5 vCPU and 1 GB; listed examples reach $192/month for 8 vCPU and 16 GB. Persistent disk costs $0.11657 per GiB/month, with extra IOPS and throughput billed separately. Egress is $0.06/GB, and a dedicated egress gateway is $15/month. Listed Northflank Cloud GPUs cost $0.80–$5.87/hour; BYOC GPU fees are $1/GB VRAM/month. Enterprise pricing is custom. The pricing page lists no annual billing option.

Northflank does not publish an open-source licence for the platform. It does offer BYOC: the platform can manage workloads in a customer’s cloud account or VPC, on-premises environment, or bare-metal infrastructure. That is a way to keep workload infrastructure under your control, but it is not the same as a documented, self-hostable control plane. Confirm deployment and operational requirements with the vendor if you need to run the platform itself without Northflank.

The homepage does not describe data-export tooling. Containerized workloads, Git repositories and standard databases give teams familiar components to move, but Northflank-specific templates, deployment settings and API calls may need replacing. Before adopting it, ask how to export database backups, logs, secrets and workload configuration, and what remains accessible if you stop using the control plane. BYOC reduces infrastructure dependence; it does not by itself remove platform or workflow lock-in.

Northflank lists autoscaling, networking, RBAC, secrets, audit logs, logs, metrics, alerts, backups, high availability and disaster recovery. It also advertises 17+ Northflank Cloud regions and GPUs on demand. Those features cover important production needs, but teams should confirm service-level objectives, backup restore procedures, regional availability, compliance coverage and API rate limits with the vendor before relying on them. The homepage currently displays a security incident notice and tells users to rotate user-defined credentials; check the latest incident updates and follow the vendor’s remediation instructions before deploying sensitive workloads.

One Reddit thread advises assuming environment variables may have been decrypted and rotating keys. That is a user warning, not a confirmed description of the incident’s scope. Other Reddit discussions show mixed views: one commenter reports an easy free deployment for a frontend and backend, while a devops thread questions whether adopting a managed platform solves the team’s underlying deployment problem. Evaluate Northflank against your existing operational model, not just its deployment interface.