Skip to main content
Choose a plan in Billing: All prices are in USD. Every plan includes Playground and API access. Pro includes more monthly usage than Plus; Ultra includes the most.

Try Limitless

Eligible new workspaces receive $0.10 in free trial credits while trial credits last. Trial credits can be used for inference at the displayed model rates. A positive available balance is required to start a request. Usage is charged after the request completes, based on the tokens used. Check your available balance in Billing; model, spending and request limits still apply. Once your available credits are exhausted, choose a plan or add prepaid credits to continue.

Start a subscription

An owner or admin chooses a plan, reviews the recurring monthly charge, and continues to Stripe to confirm payment. Subscriptions renew automatically each month until cancelled. Your plan becomes available after payment is verified. If Billing shows Checking payment, wait or select Refresh before trying another checkout.

Monthly usage

Included usage resets each billing period. Unused monthly usage does not carry forward. Purchased extra usage remains available across monthly resets until used. Your renewal date appears in Billing. The amount of work you can do depends on the models you use and the size of your requests. All plans have usage limits. Request and concurrency limits also apply; a plan does not guarantee priority capacity.

Keep working when you reach a limit

Billing shows whether usage is available or a monthly limit has been reached, along with the next reset date. You can wait for your next monthly reset, or an owner or admin can choose either option:
  • Upgrade: Move to a higher plan for more usage during the rest of the current billing period. Stripe shows the exact charge before you confirm. Your renewal date stays the same.
  • Buy extra usage: Make a one-time purchase without changing your monthly plan. Select an amount and review it in Stripe before paying. Extra usage remains available until used and does not renew automatically.
Upgrades are available for active subscriptions that are not scheduled to cancel. If payment is overdue or cancellation is scheduled, open Manage subscription first.

Manage payments and cancellation

Select Manage subscription for payment methods, invoices and cancellation. Owners and admins can manage the subscription; members can view its status. Cancel renewal anytime. Your plan remains available until the end of the paid period, and no further renewal payment is collected. Unused monthly usage ends with that period. If a renewal payment fails, a new period of usage is not added until payment succeeds. Billing shows when payment needs attention.

Pay-as-you-go API billing

Without a monthly subscription, expand Pay-as-you-go API billing in Billing to view balances, buy prepaid credits and see model rates. One-time purchases are separate from a recurring subscription. Owners and admins can add credits through Stripe when payments are configured. Choose an amount and complete payment on Stripe’s hosted checkout page. Credits appear after payment is verified. A return URL or payment screenshot does not itself confirm payment. The console shows Checking payment until it finds the confirmed purchase.

Request charges

Requests require a positive available balance when they start. They do not place an upfront monetary hold for the model’s maximum context or requested output limit. After completion, the charge uses reported input and output tokens at the model’s rates. Model context and output limits still apply. Model prices are shown in USD per million tokens. Input and output can have different rates. If usage includes cached-input tokens, those tokens use the model’s configured cached-input rate. There is no guarantee of a cache hit. DeepSeek Flash is $1 input, $0.10 cached input and $2 output per million tokens. GLM 5.3 is $3 input, $0.30 cached input and $5 output per million tokens, before applicable workspace discounts. Reasoning is part of output usage. Image input contributes to reported input usage; these routes do not add an image-generation fee. A failed request can still consume tokens. Reported usage is charged normally. A known rejection before generation has no generation charge. If the gateway observes cancellation without final usage, it makes no customer charge. Timeouts, service failures or interrupted accounting can leave usage pending reconciliation. Holds from requests admitted under the previous reservation system remain until settlement or reconciliation. Contact support with the request ID if a previous hold remains unresolved.

Spending and request limits

Requests without a positive available balance return HTTP 402 before inference. A monthly spending cap can also block a request. Rate and concurrency limits return HTTP 429; wait before retrying and respect Retry-After when present. See limits.

Programmatic billing

Customer keys with credits:read can call GET /v1/credits. Admin keys with billing:read can read /v1/organization/balance, /v1/organization/usage/completions, and /v1/organization/costs. Balances and reports belong to your Limitless organization. Messages cache-read tokens use the cached-input rate. Cache-write tokens use the normal input rate. Hosted search and fetch currently have no separate Limitless fee; reported model tokens are charged normally. Token-counting requests do not generate billable inference usage.