Pricing

    Pay for what your customers run

    No seats. No per-product plans. One price per run.

    Usage

    $0.06
    per run · LLM costs passed through at cost
    • All features included
    • Every integration and tool
    • Knowledge, memory and guardrails
    • Per-customer metering and usage API
    • White-label theming and custom domain
    Get started

    Enterprise

    Volume pricing
    for platforms running at scale
    • Dedicated infrastructure
    • Custom terms, SLA and DPA
    • Security review support
    • Named technical contact
    • Migration and exit terms in writing

    Volume pricing available from 100K runs per month.

    Definition

    What counts as a run.

    One run = one complete agent task. Model calls, tool calls, agent-to-agent hops, knowledge lookups, fallbacks and retries are all included in the same run.

    Model callsTool callsA2A hopsKnowledge lookupsFallbacksRetries
    = one run, $0.06

    Calculator

    What it costs you, and what you can charge.

    Your cost is the easy half. The half that matters is the spread you keep.

    Rate applied: $0.06 / run

    Total runs / month400,000
    You pay GTWY$24,000 / mo
    You bill your customers$100,000 / mo
    Your margin$76,000 / mo

    76% of what you bill is yours. LLM tokens are passed through at cost and are not included above.

    For comparison

    Against the work it replaces.

    A price on its own is just a number. Here is the same task, done both ways.

    A support ticket triaged by a person
    ~₹40 in staff time
    The same ticket triaged by an agent
    ₹4 per run

    Illustrative, using a typical support salary. Substitute your own numbers on the call — the ratio is what makes the case, not our arithmetic.

    LLM costs

    Tokens at cost, itemised.

    GTWY charges the run fee. Model tokens are billed at provider cost with no markup, on a separate line of every invoice — so your margin calculation doesn’t move when model prices do.

    Bring your own provider keys if you prefer. Managed keys are the default and the faster path to production.

    Storage

    The second dimension.

    If your customers use knowledge bases, stored documents and their embeddings are metered alongside runs. Two dimensions, both honest: what your customers run, and what they store.

    Platforms not using knowledge pay for runs only.

    Questions

    Before you model this out.

    What happens if a customer exceeds their limit?

    Runs stop at the cap rather than continuing and billing through it. You receive a webhook at 80% so you can raise the cap, charge for an upgrade, or let it hold.

    Can we set spend caps per customer?

    Yes. Every customer carries a default ceiling you set, overridable per customer, so one runaway agent never becomes a surprise invoice.

    How do we bill our own customers?

    Read runs per customer from the usage API and charge however your pricing works — per run, bundled into a plan, or as a premium tier. We never bill your customers directly.

    Is there a minimum commitment?

    Not on usage pricing. Enterprise agreements involve a committed volume in exchange for the volume rate and the terms that come with it.

    What’s included in enterprise?

    Volume pricing, dedicated infrastructure, custom terms, an SLA, a DPA, security review support and a named technical contact.

    Model it against your volume.

    Bring your customer count and we’ll work the margin through with you.

    Try the sandbox