Plans

Gate’s pricing has two parts that you choose separately:

Plan — which platform features your organization gets. Free, Pro, or Enterprise.
How you pay for tokens — bring your own provider keys, or top up a balance and pay through Gate. This choice is identical on every plan.

A plan does not buy tokens. It unlocks routing, security, compression, limits, audit, and team features. Tokens are always billed by your provider (OpenAI, Anthropic, and the rest) or by Gate, depending on which path you pick in part two.

Which plan is for you

Free is for trying Gate out, or for an individual running a small project on their own provider keys. The provider bills you directly; Gate is free. Basic security screening is on by default, and you can run a team of up to five people.

Pro is for teams that want the full defense layer on production traffic: transformer-based prompt-injection detection, redaction of PII / PHI / credentials on model responses, advanced compression, and spend / token / rate caps at the org, project, and key level. $20 per user per month, monthly, no annual contract.

Enterprise is for larger teams with deployment, security, and procurement requirements Gate’s self-serve plans don’t cover: custom retention and volume pricing, private cloud or VPC deployment, data isolation and export, custom security policies, dedicated onboarding, and procurement paperwork (DPA, invoicing, security review). Contact info@constellationnetwork.io.

Free

The default for every new organization. Free forever, no credit card required.

Multi-provider routing: one base URL for OpenAI, Anthropic, Gemini, Grok, and more.
Request log with cost, tokens, and latency. Conversation threading.
Immutable audit trail — every message is fingerprinted to Constellation Digital Evidence.
Response cache and prompt-cache priming.
Basic lossless compression on every request.
Gate Connect desktop app for no-code setup.

Limits and retention:

Up to 20,000 recorded requests per month.
30-day request-log retention, 90-day metrics retention.
Up to 5 members per organization (active members plus pending invitations).

Security on Free:

Basic prompt-injection screening is on by default under Policies → Free plan screening, and you can turn it off. It runs a deterministic regex scan locally on each fresh user turn with no added latency and no per-request cost. It catches the most direct attack patterns — obvious “ignore previous instructions” or role-override phrasing — and only blocks on a high multi-signal score; lower-confidence matches flag.
It does not catch paraphrased, indirect, multilingual, split-turn, or novel attacks.
Sensitivity tuning, tool-result scanning, and PII / PHI / credential redaction are Pro features.

Pro

$20 per user / month, billed per seat. The monthly total is seats × $20. Upgrade under Billing → Upgrade to Pro. Pro is monthly and self-serve — no annual contract required.

Everything in Free, plus:

Up to 200,000 recorded requests per month.
90-day request-log retention, 180-day metrics retention.
Full prompt-injection detection. The transformer cascade, sensitivity presets (Low / Medium / High), and indirect-injection scanning of tool-result content.
PII, PHI, and credential redaction on model responses, before the response returns to your agent.
Advanced compression. Cross-turn tool-result dedup and tools-schema shrinkage on top of the basic lane. Most users see 20%+ token savings per message.
Spend, token, and rate limits at the organization, project, or API-key level. When an agent hits its cap, Gate stops the requests instead of charging past it.
Unlimited seats (each one billed). The Free 5-member cap no longer applies.

How seats bill

Every active member and every pending invitation counts as a seat. Adding either requires an available seat.
Adding a seat mid-cycle is prorated — charged only for the rest of the current billing period on your next invoice.
Removing a member stops that seat’s billing at the next cycle. There is no refund or credit for the rest of the current period.
Re-adding a member within the same period is free up to the high-water mark of seats you have already paid for this period. Only growing past that watermark creates a new prorated charge.
The Billing page shows your current seat count and monthly total (N seats × $20 = $Y per month).

Limits and retention

	Free	Pro
Recorded requests per month	20,000	200,000
Request-log retention	30 days	90 days
Metrics retention	90 days	180 days
Members per organization	up to 5	unlimited (billed per seat)

Recorded requests means the number of gateway requests Gate stores in the request log and audit trail per billing cycle. Requests past the cap are still routed and scanned; they stop appearing in the log until the next cycle. Enterprise plans carry custom retention and volume terms.

Enterprise

For larger teams with deployment, security, and procurement requirements.

Custom retention periods and volume pricing.
Custom security policies.
Private cloud deployment and VPC hosting.
Data isolation and data export.
Dedicated onboarding and priority support.
Procurement: security review, DPA, invoicing.

Contact info@constellationnetwork.io to start.

Paying for model usage

This is separate from your plan and works the same on Free, Pro, and Enterprise.

Bring your own keys. Connect your provider accounts — OpenAI, Anthropic, Gemini, Grok, and more — and the provider bills you directly. Subscription tools like Claude Code keep working on your existing subscription. Gate sits in front and applies scanning, redaction, and audit on top; the model side stays exactly as it was.
Pay through Gate. No provider account or contract required. Top up a USD balance, point your tools at a single endpoint, and run any model — frontier closed-source and the best open-source side by side. Gate debits each request from your balance, you get one bill across providers, a Stripe-managed payment method, optional auto top-up, and a live balance in the dashboard.

You choose this when you create an API key and top up under Billing → Top up.

Switching plans

Free to Pro: open the Billing page, click Upgrade to Pro, and complete Stripe Checkout. The upgrade is instant; the subscription starts at your current team size (every active member is a seat).
Pro to Free: cancel under Billing → Manage subscription (the Stripe Customer Portal). Pro features stay on through the end of the current billing period. On downgrade, members beyond the Free cap of 5 stay attached (grandfathered); you just cannot invite new ones on Free until you are back under 5.
Free plan screening under Policies works on either plan — it’s an independent toggle.
Pay through Gate is also independent of plan. To turn it off, disable auto top-up and let your balance drain. Support can refund the remaining balance.

Invoices

Every charge — subscription or top-up — generates a Stripe invoice with a PDF. Find them under Billing → Manage subscription → Invoices.

FAQ

Do I need a credit card to start? No. Free is free forever and doesn’t ask for a card. You only add payment details when you upgrade to Pro or top up a Pay-through-Gate balance.
Can I use my Claude or ChatGPT subscription? Yes. Gate sits in front of the tools you already use, so you can keep your Claude Code or ChatGPT subscription, or run hundreds of models directly through Gate on pay-as-you-go.
How do spend caps work? On Pro, you can cap spend, tokens, and request rate at the organization, project, or API-key level. When an agent hits its cap, Gate stops the requests instead of your credit card.
Can I cancel anytime? Yes. Plans are monthly and self-serve. Downgrade to Free from the billing page whenever you want; your workspace keeps routing, logs, and the audit trail, and nothing is deleted.
What happens to my data when I downgrade? Your workspace keeps its routing, request log, and audit trail. Retention windows reflow to the Free limits after the next cycle unless you upgrade again.

Prompt-injection defense: what each plan’s inspection tier actually looks for.
Security: PII, PHI, and credential redaction; policy actions; retention.
Audit trail: inclusion proofs, export, retention.
How Gate works: where compression, cache, and security sit in the request flow.