Pricing
Every plan has every feature. They differ on credit, retention, and team size.
Free
free forever
Every core feature, no credit card. Local browser automation and test authoring need no account at all.
Pro
Most popularbilled monthly
Monthly credit plus pay-as-you-go, so a run never stops because the balance hit zero.
Every core feature is on the free plan, including bringing your own LLM key and running on your own infrastructure.
Compare plans
| Freefree forever | Probilled monthly | Enterprisecontract priced | |
|---|---|---|---|
| Usage | |||
| Included credit. Credit included with the plan, spent on LLM tokens, compute, and storage as you use them. This is granted to you, not billed to you. | $10 signup bonus | $60/month | Custom |
| Credit reset. How included credit arrives. It expires with its term and never carries past it; credit you buy yourself (pay-as-you-go, auto-recharge) has no term and rolls over until spent. | One-time signup bonus | Granted monthly | Custom |
| Pay as you go. Allows additional metered usage once the included credit is spent, at the plan's published rates. | Not included | Included | Custom |
| Auto-recharge. Automatically adds credit when your balance drops below the threshold you configure. | Not included | Included | Included |
| Artifact retention. How long run logs, videos, screenshots, traces, and reports remain accessible. | 30 days | 90 days | Custom |
| Usage rates. Per million tokens for LLM usage and per minute for compute. Identical on Free and Pro: a plan changes how much credit you start with, not what a token or a minute costs. | Same on every plan | Same on every plan | Custom |
| Core features | |||
| AI Action. Uses AI to understand and complete browser actions from natural-language goals. | Included | Included | Included |
| Auto-healing. Adapts test interactions when application elements or selectors change. | Included | Included | Included |
| Failure recovery. Attempts to recover from transient interaction failures during a test run. | Included | Included | Included |
| MCP access. Lets AI coding agents control a browser to inspect pages, interact with the app, verify changes, debug console and network issues, and capture evidence. | Included | Included | Included |
| AI Skills. Gives coding agents task-oriented Shiplight workflows to verify changes, review apps, write and maintain tests, and set up CI. | Included | Included | Included |
| CI integrations. Runs tests in supported CI/CD pipelines, including GitHub Actions, and uploads results to Shiplight Cloud. | Included | Included | Included |
| Action cache. Resolved locators are cached per step, so a settled step replays deterministically without calling a model. No cap on how many are stored. | Unlimited | Unlimited | Unlimited |
| Web console. Runs, tests, and recordings in the browser, with logs, screenshots, traces, and history for every run your team uploads. | Included | Included | Included |
| Advanced analytics. Run status, run duration, test status, flakiness, and failure attribution are built from the results you upload. Some usage and cost views depend on runs going through Shiplight runners or the Shiplight LLM proxy, so certain analytics may be limited when you bring your own. | Included | Included | Included |
| Bring your own LLM key. Point Shiplight at your own Anthropic, OpenAI, or Google key instead of our LLM proxy. Token usage is then billed by your provider, not by us. | Included | Included | Included |
| Run on your own infrastructure. Run the same YAML tests on your own CI runners rather than Shiplight-hosted ones. Neither our compute nor our proxy is in the path. | Included | Included | Included |
| Recordings. Session recordings kept per organization, and the maximum length of each one. | 30 recordings up to 5 min each | 150 recordings up to 15 min each | Custom |
| Ask others to record. Send a link that asks a teammate or user to capture a walkthrough recording; their recording lands in your workspace. | Not included | Included | Included |
| Email inboxes. Creates test inboxes for email-based flows such as account verification and password reset. | Unlimited | Unlimited | Unlimited |
| Administration | |||
| Organization members | 5 | 100 | Unlimited |
| Audit log | Not included | Included | Included |
| Support | |||
| Support tier | Community | Priority | Dedicated |
How usage works
Every plan includes credit, measured in dollars. Runs spend it on two things: LLM tokens when a step calls a model, and compute minutes while a runner works. After the first run, cached steps replay without calling a model, so a stable suite costs far less than its first run. And on every plan, including Free, you can skip both meters: bring your own LLM key and run tests in your own CI.
LLM models
| LiteClaude Haiku, Gemini Flash, or similar | $1.00 / 1M tokens |
| StandardClaude Sonnet, Gemini Pro, or similar | $3.00 / 1M tokens |
| ProClaude Opus, or similar | $6.00 / 1M tokens |
Compute
| shiplight-small4 vCPU, 16 GB | $0.012 / min |
| shiplight-medium8 vCPU, 32 GB | $0.024 / min |
| shiplight-large16 vCPU, 64 GB | $0.048 / min |
| shiplight-xlarge32 vCPU, 128 GB | $0.096 / min |
Authoring runs on your coding agent's own subscription, so it costs nothing here and needs no account. Running a test file needs an LLM key: ours, which spends credit, or your own.
Pricing questions
Do I need an account to try Shiplight?
No. Local MCP browser automation and YAML test authoring need no Shiplight account and no API token. Install into your coding agent and start verifying changes. An account is what gives you hosted runners, CI result uploads, dashboards, and the artifact history.
What actually consumes credit?
Two things: LLM tokens, billed per million at the model tier's rate, and compute, billed per minute while a runner is working. Cached steps replay without a model call, so a stable suite costs far less than its first run.
What happens when the credit runs out?
On Pro, pay-as-you-go takes over at the published rates, and auto-recharge can top the balance up automatically when it drops below a threshold you set. Free has no pay-as-you-go, so runs stop until the next term or an upgrade. Nothing is charged without a plan that allows it.
Can I bring my own LLM key and run on my own infrastructure?
Yes, on every plan including Free. Point Shiplight at your own Anthropic, OpenAI, or Google key and run tests on your own CI, and neither our LLM proxy nor our compute is in the path. Nothing about that is gated behind an upgrade.
Does unused credit roll over?
Included plan credit does not: it is granted for the term and expires with it, and the $10 signup bonus on Free is granted once. Credit you add yourself is different: pay-as-you-go top-ups and auto-recharge have no term, and that balance rolls over until you spend it.
How does yearly billing work?
Yearly saves 17% against paying monthly, and the full year of included credit lands on your balance upfront: you pay less and receive the same total credit. It is still term credit, so it expires at the end of the year.
Do the plans differ in what Shiplight can do?
No. AI Action, auto-healing, failure recovery, MCP access, AI Skills, and CI integrations are on Free, Pro, and Enterprise alike. Plans differ on included credit, retention, recordings, members, and pay-as-you-go, never on capability.
What does Enterprise add?
Custom credit, retention, and rates, plus SOC 2 Type II, a 99.99% uptime SLA, private cloud and VPC deployment, unlimited members, and a dedicated CSM.