# Porter

Effortless app infrastructure, your own cloud.

Source: https://wheredoirunthis.com/products/porter · JSON: https://wheredoirunthis.com/api/products/porter · Read on 2026-10-15

## At a glance

- Type: PaaS
- Kind: Hosted
- Isolation: Container
- Free tier: No
- Starting price: Not disclosed
- GPU: Yes
- Regions: 3
- Coverage: 32 of 61 assessed capabilities supported
- Agent readiness: read 11 of 11 signals, drive 10 of 11 signals

## Links

- Homepage: https://www.porter.run/
- Docs: https://docs.porter.run/getting-started/introduction
- Pricing: https://www.porter.run/pricing

## Pricing

- Model: usage-metered management fee plus separate cloud-provider bill
- Unit: requested GB RAM and vCPU; prorated to minute
- Free tier: No

| Tier | Price | Includes | Notes |
| --- | --- | --- | --- |
| Standard | $6/GB RAM/month and $13/vCPU/month | No included usage disclosed. Unlimited apps, previews, autoscaling, jobs/cron, certificates, monitoring/logging and alerts. | Both resource rates belong to Standard; not two plans or an advanced-feature unlock. Requested app resources, not node capacity, are billed. [source](https://www.porter.run/pricing) |
| Enterprise | Contact sales |  | Volume discounts starting at 40 vCPU / 80 GB RAM; premium support, advanced RBAC, SAML SSO, custom alerts and on-prem installation. [source](https://www.porter.run/pricing) |

- Overages: Usage is metered rather than an allowance plus overage. Separate cloud bill; no distinct sandbox rates disclosed.
- Card required: Not documented
- Source: https://www.porter.run/pricing

## Website design

Desktop public website · assessed 2026-10-15

Complete assessment.

Porter’s public marketing pages combine expansive black sans-serif headings, monospaced labels, yellow highlights and recurring pixel-like geometric ornaments. Pricing continues the same vocabulary with turquoise and blue plan panels; documentation switches to a dark, restrained three-column reading interface.

- Styles: Retro-computing, Playful
- Landing page: captured
- Pricing: captured
- Docs: captured

## Runtime

- **GPU & accelerators**: Yes — "Fixed NVIDIA-enabled GPU node groups; scoped to application GPU workloads, not an advertised GPU sandbox option." [source](https://docs.porter.run/other/gpu-workloads.md) (confidence: high)
- **BYOC (Bring your own cloud)**: yes — "Run and scale apps in any AWS, GCP, or Azure account. Deploy directly into your customers’ VPC." [source](https://www.porter.run/) (confidence: high)
- **Secrets management**: Yes — "Project-wide environment groups sync with cloud secret managers; sandbox env and environment-group injection supported. Workload identity available separately." [source](https://docs.porter.run/applications/configure/environment-groups.md) (confidence: high)
- **Storage options**: EFS application disks; persistent sandbox disk volumes; Porter-created or registered AWS S3 object volumes; AWS Postgres/Redis datastores — "Disk volumes have remote file operations; S3 object volumes use Mountpoint and do not allow in-place edits/appends." [source](https://docs.porter.run/sandboxes/volumes.md) (confidence: high)
- **Supported workloads**: web-services, background-workers, scheduled-jobs, pre-deploy-jobs, gpu-workloads, temporal-workers, untrusted-code-execution, agent-tools, parallel-batch-work — "Long-lived applications plus AWS sandbox capability; QM multiplayer agent harness available as cluster add-on." [source](https://docs.porter.run/sandboxes/overview.md) (confidence: high)
- **Persistence**: Yes — "Volumes live independently and keep contents across sandboxes; applications also have EFS persistent storage." [source](https://docs.porter.run/sandboxes/volumes.md) (confidence: high)
- **Data jurisdiction**: customer-controlled — "Workload region selected in customer-owned cloud; does not establish control-plane residency." [source](https://docs.porter.run/cloud-accounts/creating-a-cluster.md) (confidence: high)
- **Resource limits & quotas**: Application resources capped by underlying VM size. Sandbox Max CPU defaults to 20 cores; warm pods 0.1–16 CPU and 64–32768 MiB; single sandbox maximum 256 cores / 1 TB memory. Warm pool reserves resources; queued requests wait for capacity. [source](https://docs.porter.run/sandboxes/capacity.md)
- **SDK languages**: Python, TypeScript — "Official porter-sandbox packages; Sandbox API, not a general app-management SDK. SDK repo READMEs label rollout pre-release." [source](https://docs.porter.run/sandboxes/getting-started.md) (confidence: high)
- **Network & egress controls**: sandbox allowlists and static egress IPs — "Sandbox baseline denies private networks/metadata/control plane; enable cluster allowlist enforcement and supply hostname/IP/CIDR/service entries. Empty list denies all. Enforcement off by default; unconfigured sandbox retains internet access." [source](https://docs.porter.run/sandboxes/networking.md) (confidence: high)
- **Snapshots & forks**: Yes — "Snapshot captures a running sandbox filesystem; excludes running processes and mounted volumes. New sandboxes run their command from the beginning." [source](https://docs.porter.run/sandboxes/overview.md) (confidence: high)
- **Deployment methods**: git-push, dockerfile, container-image, buildpack, cli, sdk, api, mcp, dashboard — "Application deploy via GitHub/buildpacks/Dockerfile, registry, dashboard/CLI/MCP; Sandbox SDK and API lifecycle also available." [source](https://docs.porter.run/applications/deploy/overview.md) (confidence: high)
- **Isolation**: Container — "Kubernetes applications and isolated container sandboxes; no microVM guarantee inferred from helper repositories." [source](https://docs.porter.run/getting-started/concepts.md) (confidence: high)

## Capabilities

### AI & agents

| Capability | Status | Evidence |
| --- | --- | --- |
| Agent skills | Supported | "Quick install installs Porter’s skills: porter-deploy and porter-qm." [source](https://docs.porter.run/mcp/overview.md) |
| AI app builder | Partially supported | "Agents can publish the apps they build. Each published app runs in its own sandbox." [source](https://docs.porter.run/addons/qm.md) |
| Autonomous QA agent | Not supported | "No first-party autonomous live-app QA agent advertised." [source](https://docs.porter.run/llms.txt) |
| Built-in coding agent | Not supported | "The DevOps Agent is an AI-powered debugging assistant with read-only cluster access." [source](https://docs.porter.run/applications/observability/devops-agent.md) |
| BYOK for LLMs | Supported | "Configure provider Anthropic or AWS Bedrock, model, and your API key." [source](https://docs.porter.run/applications/observability/devops-agent.md) |
| Credential-injection proxy | Not supported | "Secrets injected into workload/env; no on-wire credential substitution keeping real secret outside runtime documented." [source](https://docs.porter.run/llms.txt) |
| LLM gateway | Not supported | "No multi-provider model gateway for user workload requests found. BYOK DevOps assistant is narrower." [source](https://docs.porter.run/llms.txt) |
| MCP client | Not supported | "No general Porter MCP-client feature documented; external agents connect to Porter as server." [source](https://docs.porter.run/llms.txt) |
| MCP-server hosting | Not supported | "No explicit customer MCP-server hosting guide or specialized hosting surface found; generic container capability not enough." [source](https://docs.porter.run/llms.txt) |
| MCP server | Supported | "Porter hosts a remote Model Context Protocol server at https://mcp.porter.run." [source](https://docs.porter.run/mcp/overview.md) |
| Model inference API (first-party) | Not supported | "GPU inference workloads supported, but Porter does not offer own model-catalogue inference API." [source](https://docs.porter.run/llms.txt) |
| Preinstalled coding agent | Not supported | "MCP connects external harnesses; no Claude Code/Codex/etc. preinstalled sandbox session claimed." [source](https://docs.porter.run/llms.txt) |

### Networking & delivery

| Capability | Status | Evidence |
| --- | --- | --- |
| Automatic HTTPS | Supported | "Automatic Let’s Encrypt SSL certificate issuance and renewal." [source](https://docs.porter.run/applications/configure/custom-domains.md) |
| Custom domains | Supported | "You bring a domain you own; Porter provisions a load balancer and a wildcard HTTPS certificate." [source](https://docs.porter.run/sandboxes/networking.md) |
| Direct public IPv6 | Not supported | "No per-workload stable public IPv6/raw TCP-UDP endpoint documented." [source](https://docs.porter.run/llms.txt) |
| Edge KV store | Not supported | "No colocated edge KV primitive; workload clusters are customer cloud Kubernetes." [source](https://docs.porter.run/llms.txt) |
| Egress controls | Supported | "Pass egress when creating a sandbox and it can only reach the destinations it lists; all other outbound traffic is blocked." [source](https://docs.porter.run/sandboxes/networking.md) |
| Email send / receive | Not supported | "No first-party workload email primitive found; notification emails are not workload email service." [source](https://docs.porter.run/llms.txt) |
| Public preview URLs | Supported | "Deploy your first application from sign-up to a live public URL." [source](https://docs.porter.run/getting-started/quickstart.md) |
| Private networking | Supported | "Porter peers the datastore VPC to your cluster VPC and ensures traffic flows exclusively through private subnets." [source](https://docs.porter.run/addons/datastores.md) |
| Built-in reverse proxy | Supported | "Customize NGINX ingress annotations including timeouts, body size, and WebSocket support." [source](https://docs.porter.run/applications/configure/advanced-networking.md) |

### Organisation & security

| Capability | Status | Evidence |
| --- | --- | --- |
| Built-in auth | Partially supported | "Each pod assumes a cloud identity at runtime via the cloud provider’s native workload-identity mechanism." [source](https://docs.porter.run/applications/configure/secure-cloud-access.md) |
| BYOC | Supported | "Run and scale apps in any AWS, GCP, or Azure account." [source](https://www.porter.run/) |
| Deploy into customers' VPCs | Supported | "Deploy directly into your customers’ VPC." [source](https://www.porter.run/) |
| Secrets management | Supported | "Share environment variables and secrets across applications using project-wide groups synced to cloud secret managers." [source](https://docs.porter.run/applications/configure/environment-groups.md) |
| Self-host / OSS core | Not supported | "Public organization repositories include SDKs, charts, API contracts and Switchboard; no evidenced self-hostable Porter PaaS core." [source](https://github.com/porter-dev) |
| Sharing | Supported | "Public networking: Anyone on the internet can reach sandboxes at a hostname under your domain." [source](https://docs.porter.run/sandboxes/networking.md) |
| SSO / SAML | Partially supported | "Click Request under SSO provisioning to reach out to support." [source](https://docs.porter.run/security-and-compliance/role-based-access-control.md) |
| Teams | Supported | "Assign Admin, Developer, or Viewer roles to members in a Porter project." [source](https://docs.porter.run/security-and-compliance/role-based-access-control.md) |
| Workload identity | Supported | "Porter services authenticate without static access keys: EKS Pod Identity, Azure Workload Identity and GCP service-account bindings." [source](https://docs.porter.run/applications/configure/secure-cloud-access.md) |

### Developer access

| Capability | Status | Evidence |
| --- | --- | --- |
| CLI | Supported | "Install the Porter CLI via Homebrew, shell script, or manual binary download." [source](https://docs.porter.run/cli/installation.md) |
| E2B-compatible API | Not supported | "Official SDKs use Porter Sandbox API; no drop-in E2B compatibility claim found." [source](https://docs.porter.run/llms.txt) |
| Filesystem API | Supported | "Browse, read, write, upload, download and move disk-volume files from SDK, CLI or dashboard." [source](https://docs.porter.run/sandboxes/volumes.md) |
| Git deploy | Supported | "Connect a GitHub repository for automatic builds and deployments with CI/CD triggers." [source](https://docs.porter.run/applications/deploy/deploy-from-github-repo.md) |
| HTTP data API | Not supported | "No first-party data-query REST/GraphQL interface advertised; Sandbox/control-plane API is not a data API." [source](https://docs.porter.run/llms.txt) |
| OpenAPI spec | Partially supported | "Everything under porter_sandbox/ is generated from the Porter Sandbox OpenAPI spec." [source](https://github.com/porter-dev/porter-python-sdk) |
| Private container registry | Supported | "List project-linked container registries and push images without running docker login." [source](https://docs.porter.run/standard/cli/command-reference/porter-registry.md) |
| Public API | Supported | "SDK connects to in-cluster Sandbox API automatically; external callers use PORTER_SANDBOX_API_KEY and PORTER_CLUSTER_ID." [source](https://docs.porter.run/sandboxes/getting-started.md) |
| SDK | Supported | "pip install porter-sandbox; npm install porter-sandbox." [source](https://docs.porter.run/sandboxes/getting-started.md) |
| SSH access | Not supported | "CLI run/exec opens application shell; no direct SSH endpoint documented." [source](https://docs.porter.run/standard/cli/command-reference/porter-app.md) |

### Compute & storage

| Capability | Status | Evidence |
| --- | --- | --- |
| Cloud-init / ignition config | Not supported | "No user launch-time cloud-init/ignition interface documented." [source](https://docs.porter.run/llms.txt) |
| Desktop / computer-use sandbox | Not supported | "No desktop, VNC or computer-use environment documented." [source](https://docs.porter.run/llms.txt) |
| GPU compute | Supported | "Create fixed GPU node groups with NVIDIA-enabled instances." [source](https://docs.porter.run/other/gpu-workloads.md) |
| Live-state (memory) snapshots | Not supported | "A snapshot holds files only, not running processes or mounted volumes." [source](https://docs.porter.run/sandboxes/overview.md) |
| Multi-region | Supported | "Select region, machine type, and initial instance count in your AWS, GCP, or Azure cloud account." [source](https://docs.porter.run/cloud-accounts/creating-a-cluster.md) |
| Object storage | Supported | "Create a new bucket or register an existing bucket through Add-ons > Object storage." [source](https://docs.porter.run/sandboxes/volumes.md) |
| Persistent disk | Supported | "A volume lives independently and keeps its contents across sandboxes." [source](https://docs.porter.run/sandboxes/volumes.md) |
| Sandbox templates / custom images | Supported | "Snapshot an environment after installing dependencies or cloning a repository, then start many sandboxes from that point." [source](https://docs.porter.run/sandboxes/overview.md) |
| Scale to zero | Partially supported | "When a service is put to sleep, it will be stopped and no instances will run." [source](https://docs.porter.run/applications/configure/basic-configuration.md) |
| Selectable isolation runtime | Not supported | "No documented user-selectable container versus gVisor versus microVM isolation modes; public gVisor fork alone is insufficient." [source](https://docs.porter.run/llms.txt) |
| Snapshots / forking | Supported | "A snapshot captures a running sandbox’s filesystem so you can start new sandboxes from it later." [source](https://docs.porter.run/sandboxes/overview.md) |

### Data

| Capability | Status | Evidence |
| --- | --- | --- |
| Database branching | Supported | "Aurora clusters support cloning an existing cluster using fast-cloning for experiments or recovery." [source](https://docs.porter.run/addons/datastores.md) |
| Ephemeral instant DB (no-signup) | Not supported | "Datastore creation requires existing Porter project/cloud cluster, not anonymous no-signup DB." [source](https://docs.porter.run/llms.txt) |
| First-party database | Supported | "Provision managed Postgres and Redis databases on AWS." [source](https://docs.porter.run/addons/datastores.md) |
| Vector search | Not supported | "No first-class vector search/index API advertised in datastore docs." [source](https://docs.porter.run/llms.txt) |

### Platform & operations

| Capability | Status | Evidence |
| --- | --- | --- |
| Managed CI/CD runners | Not supported | "GitHub Actions builds use existing CI; no separately offered managed CI runner compute documented." [source](https://docs.porter.run/llms.txt) |
| Built-in observability | Supported | "CPU, RAM, network usage and throughput metrics with 14-day retention via in-cluster Prometheus." [source](https://docs.porter.run/applications/observability/monitoring.md) |
| Preview environments | Supported | "Created automatically for each pull request in isolation and destroyed once closed or merged; addons can be created alongside." [source](https://docs.porter.run/preview-environments/overview.md) |
| Scheduled tasks | Supported | "Job services include cron schedules, timeouts, concurrency controls, and resource limits." [source](https://docs.porter.run/applications/configuration-as-code/services/job-service.md) |
| Project template marketplace | Not supported | "No project-starter template marketplace found; preview templates and runtime snapshots are different." [source](https://docs.porter.run/llms.txt) |
| Durable workflows | Partially supported | "Scale Temporal workers based on task queue backlog with a Temporal Cloud API key." [source](https://docs.porter.run/applications/configure/temporal-autoscaling.md) |

## Agent readiness

### Read: 11 of 11 signals present

11/11 signals (8 full, 3 partial)

- Llms txt: pass — "Substantive docs index https://docs.porter.run/llms.txt; apex llms redirects readers there."
- Llms full txt: pass — "https://docs.porter.run/llms-full.txt 200; 657403 bytes of complete modular docs retained as bounded pointer plus individual scored pages."
- Robots AI policy: partial — "https://www.porter.run/robots.txt allows * except /draft/; no named AI UA policy."
- JSON LD: pass — "Homepage SoftwareApplication structured data with description and offers. Offer expiry 2025-12-31 is stale, not current free-tier evidence."
- Sitemap: pass — "Non-trivial apex and docs XML sitemap fetched successfully."
- Meta quality: pass — "Homepage title, descriptive description and OpenGraph title/description available in server HTML."
- Markdown negotiation: partial — "Homepage Accept: text/markdown stays text/html; all indexed docs .md variants fetch clean markdown."
- Content efficiency: partial — "Homepage readable-text 3564 bytes / 229440 HTML bytes; script/style excluded and whitespace normalized; between 1% and 5%. See homepage-evidence.json."
- Docs discoverable: pass — "Homepage primary Resources navigation links docs.porter.run/getting-started/introduction."
- Pricing machine readable: pass — "Standard per-resource and Enterprise tiers present in parseable server-rendered pricing HTML."
- Anti bot: pass — "All scored homepage/pricing/docs sources fetched without public-content challenge. Dashboard login redirect is auth, not public-content anti-bot."

### Drive: 10 of 11 signals present

10/11 signals (6 full, 4 partial)

- Public API: pass — "Sandbox API lifecycle fully documented via SDK references, with external token/cluster and base URL configuration. App operations additionally via MCP/CLI."
- Openapi spec: partial — "Official Python/TypeScript README states generated from Porter Sandbox OpenAPI spec. No downloadable spec URL found; common apex/docs paths 404 and dashboard spec routes render login."
- MCP server: pass — "Documented operational https://mcp.porter.run endpoint with OAuth and tools reference; docs .well-known/mcp also returns anonymous Mintlify docs endpoint."
- SDK packages: pass — "Official Python pip porter-sandbox and TypeScript npm porter-sandbox documented; rollout labeled pre-release."
- Auth docs: pass — "Sandbox API token from Settings > API tokens; issuance requires admin and token developer permission. MCP browser OAuth and CLI token login separately documented."
- Agent self serve creds: partial — "Dashboard-issued API token and browser-OAuth MCP require human sign-in; no unauthenticated programmatic credential issuance documented."
- Webhooks: fail — "No publicly documented outbound webhook/signing workflow in current docs index. API-contracts webhook protobuf is not public webhook setup documentation."
- CLI tool: pass — "Official CLI installation/reference and Sandbox CLI lifecycle documented."
- Error model: partial — "SDK error classes document message/status_code/parsed body, authentication/not-found/rate/server errors; no common JSON wire-error schema published."
- Rate limit docs: partial — "RateLimitError documented when API returns rate-limit response; no numeric limits or response headers disclosed."
- Agent config: pass — "Apex agents.md plus docs agent instructions, installable porter-deploy/porter-qm skills and installer https://agents.porter.run; installer fetched but never executed."

## Coding harnesses

- Claude Code (Anthropic) — "claude mcp add --transport http porter https://mcp.porter.run" [source](https://docs.porter.run/mcp/overview.md)
- Codex CLI (OpenAI) — "codex mcp add porter --url https://mcp.porter.run" [source](https://docs.porter.run/mcp/overview.md)
- OpenCode (opencode) — "opencode mcp add" [source](https://docs.porter.run/mcp/overview.md)
- Cursor (Anysphere) — "Create or edit .cursor/mcp.json and connect Porter in Tools & MCP." [source](https://docs.porter.run/mcp/overview.md)

## Compliance & regions

- Regions: AWS, Google Cloud, Microsoft Azure
- Regions note: Supported clouds, not Porter-operated workload regions. Customer chooses region when provisioning its cloud cluster; no exhaustive supported-region list. Sandboxes/datastores presently AWS-only; no public control-plane region disclosed.

## Company & presence

- Headquarters: New York City, NY
- Funding: Series A, $20000000, 2026-01-27
- Privacy policy: https://docs.porter.run/terms-and-privacy/privacy-policy
- Terms of service: https://docs.porter.run/terms-and-privacy/terms-of-service
- GitHub: https://github.com/porter-dev
- Y Combinator: https://www.ycombinator.com/companies/porter
- Blog: https://www.porter.run/blog
- Changelog: https://www.porter.run/changelog

## Nearby in the atlas

- [Aptible](/products/aptible.md)
- [Canine](/products/canine.md)
- [DigitalOcean App Platform](/products/digitalocean-app-platform.md)
- [Fly.io](/products/fly-io.md)
- [here.now](/products/here-now.md)
- [INK](/products/ink.md)

## Method

Facts come from the product's own public pages. Each claim carries its evidence: a quote and the page it came from. We count capabilities rather than grade products, and "Not assessed" means we have not checked, never that the product lacks it.