Simple pricing
Plans for every stage
Same powerful AI API. Different scales. Choose what fits your needs.
What you are charged for
What the request cost to serve — not how large it was. The two are usually very different, and the difference is in your favour: a long prompt that repeats context the model has already seen is served far more cheaply than its token count suggests, and that saving is passed on rather than kept.
- Metered is not billed. Your usage page shows both. The metered figure is the size of the request as the model counted it; the billed figure is what left your balance. A large agent turn can meter tens of thousands of tokens and bill a handful.
- Every request is itemised. Your usage page shows each one, what it metered, what it billed, and whether those counts came from the model or were estimated.
- Nothing else. No per-request fee, no per-key fee, no seat count, and no charge for a request that did not run.
Starter
For a first project or an evaluation. The smallest package we sell.
USD 17.5
35,000,000 tokens
- Great for testing and small projects
- Access to all supported models
- Same reliable API
- Developer-friendly
Growth
A working developer's month of day-to-day use.
USD 50
100,000,000 tokens
- Ideal for regular use and development
- Access to all supported models
- Same reliable API
- Scale your ideas
★ Most Popular
Scale
A small team, or one heavy agent workload.
USD 125
250,000,000 tokens
- Perfect for small teams or one heavy agent
- Access to all supported models
- High-throughput workloads
- Built for scale
Volume
Continuous pipelines, batch jobs and high-volume agents.
USD 250
500,000,000 tokens
- Built for continuous pipelines and batch jobs
- Access to all supported models
- Optimized for high volume
- Same reliable API
Custom
Need something different?
Let's talk
Any volume
- Custom token amounts
- Tailored to your needs
- Ideal for large-scale or unique use cases
- Flexible options
Everything visible, nothing to reconcile
One balance, one dashboard, and a live view of exactly what you hold and exactly what you have spent — kept current with every request you make.
Your balance is a single number, and it is always the current one. No statements to reconcile, no invoices to match against a spreadsheet, and no month-end arithmetic to work out what you actually used.
Every request is listed individually — the tokens it read, the tokens it wrote, what it drew from your balance, and how long it took. You can see where your spend goes rather than taking a summary on trust.
What you have, what you have used, and what each request cost are all on one screen. Whether you are running a single test or a fleet of agents, the answer to "where do I stand" is one glance away.
Ready to grow with you
Start where you are and add credits as your traffic does — no renegotiation, no re-provisioning, and nothing to interrupt what you are already running.
Adding credits takes moments and applies straight away. Buy what you need rather than what a tier allows, and spend it across any model, any key and any project you run.
Credits are yours to spend at your own pace. Nothing expires while your account is open, and nothing is charged while you are not using the service — so a quiet week costs you nothing and a busy one is already covered.
Move from a first integration to production traffic without a code change, a new agreement or a migration. The service grows to fit the workload, not the other way round.
Buying more
The packages are the fast path. Anything larger, or an unusual shape of traffic, is priced on what you actually need rather than rounded up to the nearest tier — and because you are buying tokens rather than a subscription, the balance you already have carries over untouched.
Ready when you are
An account takes a moment and needs no card. Your keys are ready the moment you sign in and the balance appears as soon as it is funded — nothing to configure first, and nothing to cancel later.