Plans and billing
Plans, spend budgets, usage packs, agent tool pricing, and how billing works.
Plans are monthly spend budgets, not fixed token allowances. Inference is metered at the per-model rates on the Models page and deducted from your budget. Usage packs top that budget up without a subscription.
| Plan | Monthly | Yearly | Spend budget |
|---|---|---|---|
| Pro | $20 | $192 (save 20%) | $40 per month |
| Max | $60 | $576 (save 20%) | $120 per month |
| Team | $500 | $4,800 (save 20%) | $1,000 pooled per month, coming soon |
| Enterprise | Custom | Custom | Custom |
In short: Pro costs $20 per month, Max costs $60 per month, Team costs $500 per month and is coming soon, and Enterprise pricing is custom.
How the spend budget works
- Each plan grants a monthly spend budget:
$40for Pro,$120for Max, and$1,000pooled for a Team. The budget resets on the same calendar cadence every month. Subscribers who joined before the October 2026 change keep their original budget ($60for Pro,$300for Max) while their subscription stays active, including when they switch between Pro and Max. - Inference and web search draw from the same budget.
- Token totals are estimates for DeepSeek V4.1 Flash, the lowest-priced model, with typical cached agent usage. They are not a cap. Other models use the budget faster, and actual totals vary by prompt shape, output length, and cache reads.
- Usage packs add spend that expires 90 days after purchase. They can be bought with or without a subscription.
Pro covers up to 1B tokens per month and Max up to 3B tokens per month.
Usage packs
| Pack | How far it goes |
|---|---|
$5 | Experiments and prompt tuning. |
$10 | Short build sessions. |
$20 | Evals, agents, or a busy launch week. |
$50 | Heavier agent runs and team testing. |
$100 | Large batches and long agent loops. |
Usage packs cost $5, $10, $20, $50, or $100 as a one-time purchase, and each expires 90 days after purchase. Packs are non-recurring and never auto-charge. Purchased spend is added to your balance and consumed only after a plan's monthly usage is used up or if no plan is active.
Tool pricing
| Tool | Price | Billing |
|---|---|---|
| Web Search | $0.02 | Per search, from the same spend balance as inference. |
| Web Search Pro | $0.20 | Per deep search, from the same spend balance as inference. |
| Fetch | $0.005 | Per URL fetch, from the same spend balance as inference. |
| Fetch Pro | $0.02 | Per JavaScript-rendered URL fetch, from the same spend balance as inference. |
Agent tools are MCP tools that can be configured in your harness. Their cost is deducted from the plan or usage-pack balance and is not included in token estimates.
Billing details
- Yearly plans are billed annually and provide the same monthly spend budget as monthly plans. Choosing yearly saves 20% (about two months free).
- Cancellation takes effect at the end of the current billing period. Access remains active until then, and there are no prorated refunds.
- Taxes and currency are handled by the payment provider at checkout.
Per-key controls
Account spend is separate from per-key controls. Each API key can carry a spend quota, an expiry, region selection, and can be revoked at any time. See Key controls.
Enterprise
For team and organization workflows, shared API-key planning, a custom usage plan, custom SLA, direct support, security and procurement support, and on-prem deployment, see Enterprise or contact sales.