Northflank
Also known as: northflank.com, Northflank Sandboxes, Northflank Cloud Agents
Northflank is a cloud platform that runs AI agents and the code they generate in microVM sandboxes and cloud coding harnesses, alongside services, jobs and databases, on Northflank Cloud or in the buyer's own cloud.
Northflank sells a runtime platform for software that agents build and operate. It runs services, jobs, managed databases, GPU workloads and preview environments on Kubernetes, and its agent products are Sandboxes, isolated microVMs for agents and untrusted code, and Cloud agents, which run Claude Code, Codex, OpenCode, Pi or Cursor in the cloud with a persistent workspace. Northflank Ltd is based in London, and one control plane drives both Northflank Cloud and clusters in the customer's own cloud account.
A CPU sandbox runs in a microVM on Cloud Hypervisor with its own guest kernel, and a GPU sandbox runs under gVisor. Sandboxes are ready in under a second from a cached image, scale to 288 vCPU and 2 TB of memory, run with no time limit, pause to stop compute charges, and more than 50,000 can run at once.
Volumes reach 1 to 5 GB/s and 64 TB and can be shared across sandboxes, S3 buckets mount through FUSE, secrets can be added to outbound requests at egress so they never enter the sandbox, and network policies, mTLS and Tailscale control traffic.
Cloud harnesses keep a 5 to 100 GB workspace through restarts, push only to their own branch and expose live previews, and workflows carry agent built changes through preview, staging and production with approval steps along the way.
Pricing is per second with no seat fees. The free Sandbox plan covers up to $20 a month of resources, pay as you go starts at $0.01667 per vCPU hour and $0.00833 per GB hour of memory, BYOC costs $0.02083 per managed node vCPU hour with the cloud provider billing the infrastructure, and Enterprise adds a platform fee. Northflank holds SOC 2 Type II and ISO 27001 and supports HIPAA, with fine grained roles on paid plans and SAML, SCIM and audit logging on Enterprise. Projects run in more than 17 regions or in the customer's AWS, Google Cloud, Azure, Oracle, CoreWeave or Nebius account, and Enterprise can run the control plane itself inside the customer's network, air gapped if needed.
Northflank fits teams that want agent sandboxes, coding agents and the production services those agents ship on one platform, under one set of roles, audit logs and release workflows, or that need sandboxes with GPUs or inside their own cloud. It fits less well for a buyer who wants a ready integration catalog for agents, or only a narrow sandbox API without a wider deployment platform to learn.
Vendor details
Canonical URL
https://northflank.com
Category
Agent infrastructure
Funding status
Northflank is independent. Northflank Ltd is registered in London as company 11918540.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
A REST API, a CLI, JavaScript and Python clients and a hosted MCP server for Claude Code, Codex, Cursor and OpenCode drive the platform. Builds and workflows connect to GitHub, GitLab, Bitbucket, self hosted GitLab, GitHub Enterprise and Azure DevOps, workloads reach AWS, Google Cloud and Azure through workload identity, notifications go to Slack, email, Teams, Discord and webhooks, logs stream to third party observability tools, and Tailscale joins sandboxes to private networks.
In practice
Your product runs a separate agent for every customer and the tenants must stay apart. Northflank gives each agent its own microVM sandbox with a private endpoint, and adds its model key at egress so the key never sits inside the sandbox.
Your team wants Claude Code working on a fix after everyone logs off. A Northflank cloud harness keeps the session running in a microVM, pushes only to its own branch and puts a live preview on a public URL for review in the morning.
You need 50,000 reinforcement learning environments for one training run. Fan them out as sandboxes from one image with a shared dataset volume, billed by the second and deleted when the batch ends.
Sources & related URLs
Research sources
Agentic Index coverage score
9.0 / 14 capabilities · 64%
| Integrations & Tool Calling | Partial |
|---|---|
|
Agents reach outside systems through a short, fixed set of built in routes. A cloud harness pushes to Git through a proxy that limits its credentials to the harness branch and refreshes them automatically, workload identity issues short lived tokens for AWS, Google Cloud and Azure, and workflow message steps post to Slack, a webhook or a pull request comment. Any other API key can be added to outbound requests at egress so the agent's code never holds it, but the tools that use it are the customer's own code, and agents get no integration catalog or tool framework. Sourcenorthflank.com/product/sandboxesread 2026-11-05 |
|
| Workflow Orchestration | Partial |
|
Workflows chain builds, deployments, job runs, backups, conditions, approvals and messages in a visual editor or as JSON, with sequential, parallel and loop control and child workflows that can wait for completion, and the cloud agent loop carries agent built changes through previews, staging and production this way. A job can run an agent and the workflow waits for its result. These steps orchestrate releases and infrastructure around the agents, while the agent's own reasoning steps, hand offs and retries stay inside the customer's harness or code. Sourcenorthflank.com/docs/v1/application/release/configure-workflowsread 2026-11-05 |
|
| Knowledge Grounding & RAG | Not documented |
|
No retrieval index or knowledge base over the customer's content is part of the platform. Volumes, S3 buckets mounted through FUSE and managed Postgres, MySQL, MongoDB or Redis hold data that an agent's code can read and write, and any vector store or retrieval layer is something the customer deploys on top of them. Sourcenorthflank.com/product/sandboxesread 2026-11-05 |
|
| Human Oversight & Guardrails | Partial |
|
Release workflows can wait on people. An approval step holds a workflow until the required number of approvals is given from its run page, so an agent built change can be stopped before it reaches staging or production, and cloud harness pushes are confined to the harness's own branch. Nothing holds an agent's direct API, CLI or MCP actions for review, since those run with whatever the token's role permits. Sourcenorthflank.com/docs/v1/application/release/configure-workflowsread 2026-11-05 |
|
| Security, Identity & Governance | Full |
|
Northflank holds SOC 2 Type II and ISO 27001 and supports HIPAA, with reports in its trust center. Fine grained roles grant create, read, update and delete permissions per resource type, can be limited to or excluded from projects, and drive API tokens, which inherit their role's permissions. Owner, admin and default roles come preset, organization roles span teams and can follow directory groups, and Enterprise adds SAML and OIDC single sign on with SCIM and just in time provisioning. Team policies limit which coding agents may run and can keep a harness private to its creator. Sourcenorthflank.com/docs/v1/application/secure/use-role-based-access-controlread 2026-11-05 |
|
| Observability & Auditability | Full |
|
Every sandbox, job and harness streams logs and reports CPU, memory and network metrics through the dashboard and API, and workflow, job and template runs keep their run details. On Enterprise, audit logs record every action from the UI, API and CLI, templates, release flows, Git events and the system, with the acting user, origin, affected resources, a before and after diff and the parent events that caused it, at organization, team, project and resource level, and forward to S3. Log sinks, alerts and notifications to Slack, email, Teams or webhooks come on paid plans. An agent's prompts and tool calls inside its harness are not traced. Sourcenorthflank.com/docs/v1/application/observe/audit-logsread 2026-11-05 |
|
| Memory & State Persistence | Partial |
|
State outlives a single run. A cloud harness keeps its home directory on a 5 to 100 GB volume through restarts and pauses, sandboxes keep attached volumes across pause and resume, and S3 buckets mounted through FUSE keep data after a sandbox is deleted. Northflank's stateful agent example gives each agent a managed Postgres database for memory and resumes its session against it. That memory is a database and volume the customer's own agent code writes and reads, with no memory layer of stated scope and lifetime managed by the platform. Sourcenorthflank.com/product/sandboxesread 2026-11-05 |
|
| Deployment & Data Residency | Full |
|
Projects run in one of more than 17 Northflank Cloud regions, chosen at creation and fixed afterwards, with GPU sandboxes in GPU enabled regions. BYOC runs the same sandboxes and harnesses in the customer's own AWS, Google Cloud, Azure, Oracle, Civo, CoreWeave or Nebius account, with Kata microVM support for harnesses, and Northflank also manages bare metal, OpenShift and VMware clusters. Enterprise can run the control plane itself inside the customer's infrastructure, air gapped if needed, so no platform data leaves the network. Sourcenorthflank.com/docs/v1/application/sandboxes/sandboxes-on-northflank-cloudread 2026-11-05 |
|
| Prebuilt Agents / Templates / Packs | Partial |
|
Cloud harnesses come preset for Claude Code, Codex, OpenCode, Pi and Cursor, or start empty for other tooling, and a Northflank skill and plugin teach coding agents the platform. Stack templates deploy common applications in one step, and templates and preview blueprints package infrastructure and releases for reuse. These are configured third party agents and building blocks a developer assembles, not prebuilt agents of Northflank's own. Sourcenorthflank.com/product/cloud-agentsread 2026-11-05 |
|
| Triggers & Channel Coverage | Full |
|
Work starts from Git events, schedules and webhooks. Workflows and preview blueprints run on a push, a pull request, a tagged release, a completed check suite, a pull request label, a cron schedule no more often than every 10 minutes, or a GET or POST to a webhook URL, with the branch, commit and pull request passed into the run. Jobs also run on cron schedules or on demand through the API, so an agent job can start on a timer or when another system calls it. Sourcenorthflank.com/docs/v1/application/release/configure-workflowsread 2026-11-05 |
|
| Model Flexibility & Routing | Full |
|
The customer chooses the agent and the model behind it. A harness runs Claude Code, Codex, OpenCode, Pi or Cursor signed in with the customer's own API key or account, or starts empty for other tooling, and team policies decide which of those agents a team may run. Models can also be self hosted on GPUs from L4 to B200 for inference beside the agents, and Northflank routes no model traffic of its own. Sourcenorthflank.com/product/cloud-agentsread 2026-11-05 |
|
| APIs / SDKs / MCP Extensibility | Full |
|
Every resource is reachable from outside through a REST API, a CLI, and JavaScript and Python clients, with endpoints for sandboxes, harnesses, jobs, workflows, templates, secrets, volumes and audit logs, plus exec, file copy, log tailing and metrics. A hosted MCP server takes a team API token as a bearer header and lets Claude Code, Codex, Cursor or OpenCode deploy services, manage jobs and databases and inspect resources. Templates and GitOps define the same resources as code. Sourcenorthflank.com/docs/v1/api/introductionread 2026-11-05 |
|
| Testing, Debugging & Optimization | Not documented |
|
There is no scoring, test case or quality gate for an agent's behavior. Sandboxes run reinforcement learning rollouts and evaluations at more than 50,000 environments in parallel from one image with shared volumes, but the rewards and scores come from the customer's own code. Sourcenorthflank.com/product/sandboxesread 2026-11-05 |
|
| Browser / Computer-use | Partial |
|
Browser agents run on a browser image the customer picks, such as headless Chromium from browserless, with one sandbox per session that the agent drives over the private network through the DevTools protocol, behind egress rules set by domain. Northflank supplies the isolated compute and networking for it, not a browser or desktop of its own. Sourcenorthflank.com/product/sandboxesread 2026-11-05 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Pricing
Free Sandbox plan, then pay as you go from $0.01667 per vCPU hour and $0.00833 per GB hour of memory.
Per second vCPU and memory, plus volumes, egress and GPUs
Included quota
The free Sandbox plan covers 2 services, 2 jobs and 1 database with up to $20 a month of resources on Northflank Cloud, plus one BYOC cluster of up to 12 vCPU and 48 GB. A harness on the free plan can use up to 1 vCPU and 1 GB.
What is public
Northflank publishes every usage rate, the price of each compute plan, GPU hourly rates, the BYOC fee per managed vCPU and a full plan comparison. The Enterprise platform fee is not published.
Billing mechanics
Compute bills per second on the plan size a workload runs, with memory billed separately on Northflank Cloud and volumes billed per GiB month whether a workload runs or is paused. On BYOC only the managed node vCPU is billed, and the cloud provider bills the infrastructure directly. Invoices are monthly in US dollars.
Cost watchouts
Persistent volumes bill per GiB month even while a sandbox is paused, network egress costs $0.06 per GB, disk IOPS above 3,000 and throughput above 140 MiB/s are charged extra, and a dedicated egress gateway is $15 a month. On BYOC the cloud provider bills the infrastructure on top of Northflank's fee.
Variable cost rationale
Cost follows the plan size and seconds each sandbox or harness runs, plus volumes, egress and GPUs, so a large or long lived fleet costs many times a small one.
Additional watchouts
Fine grained roles start on pay as you go, while SAML, SCIM and audit logging are Enterprise only. Prices exclude VAT and GST.
Overage / add-ons
Usage is billed at the end of each monthly cycle, prorated to the second, in US dollars.
Sales call required
Mixed (some tiers require a call)
Free / trial
Free Sandbox plan, no card needed, with up to $20 a month of resources
Lowest paid plan
Pay as you go usage from $0.01667 per vCPU hour
Commercial notes
Northflank Ltd is registered in London. Enterprise adds a 99.99% uptime SLA, a dedicated Slack channel with Northflank's engineers, white labeling and usage discounts at spend thresholds.
Key ambiguities
The Enterprise platform fee, and with it SAML, SCIM, audit logging and the air gapped control plane, is priced on request.
Missing data
The Enterprise platform fee and its support tiers have no public price.
Related vendors
- AgentOps — Agent observability and debugging platform: open source SDKs trace…
- AgentPhone — AgentPhone gives the AI agents a team builds their own US or…
- Agno — Python agent framework and AgentOS runtime (formerly Phidata) for…
- AIsa — Resource and payment gateway for AI agents, with one key to 110+…
- Albato — Albato is a no code integration platform that sells an embedded,…
- AlphaBitCore — AI control plane for regulated financial firms: one gateway enforces…
Alternatives to Northflank
The closest documented capability profiles to Northflank among agent infrastructure platforms tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Cosmonic9.0 / 14Fuller documented coverage on Integrations & Tool Calling
- Modal8.0 / 14A lighter documented profile than Northflank
- Blaxel8.5 / 14Fuller documented coverage on Integrations & Tool Calling
- Composio9.5 / 14Fuller documented coverage on Integrations & Tool Calling and Human Oversight & Guardrails
- Firecrawl8.5 / 14Fuller documented coverage on Browser & Computer Use
- Albato10.0 / 14Adds documented Testing, Debugging & Optimization
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded