Plans & billing

Where to find it

Billing changes (subscribe, change plan, top up) are made by the owner of the active workspace. A member working in a shared workspace switches to their personal workspace to manage their own billing.

Plans and prices

There are five plans: Free, Starter, Pro, Team, and Enterprise. A paid plan's price is its base price plus the price of the monthly credit tier you choose (see Credit tiers). Prices are in US dollars.

Enterprise is not self-serve: choosing it opens a contact form instead of checkout, on both the cloud and the Community Edition.

What each plan includes

The features gated by plan on the cloud (vector search and the Browser Agent from Pro, your own LLM keys from Team) are never gated on a self-hosted install.

Storage is shared per account

The storage allowance belongs to the account, not to each workspace. Files stored in every workspace you own count against one pool. When the pool is full, every one of those workspaces refuses new files, even one that has stored nothing itself. Check usage in Settings > Storage, and see Files & storage.

Credit tiers

On a paid plan you choose how many credits you get each month. The base price includes 10,000 credits; each larger tier adds a fixed amount to it, and the price per extra credit drops as the tier rises. Team has its own price curve.

For example, Pro with 25,000 credits billed monthly is $24 + $15 = $39 per month; billed yearly it is $19 + $15 = $34 per month.

Top-ups (pay as you go)

Independently of your subscription, the workspace owner can buy a one-time credit top-up from Settings > Quota & Usage. Checkout is handled by Stripe, and the credits arrive once the payment is confirmed.

That is $1.25 per 1,000 credits, deliberately more than what each larger Starter or Pro credit tier costs per credit it adds, so on those plans a bigger tier is the cheaper way to buy credits. Top-ups are meant for occasional overflow. Top-up credits never expire.

What a credit is worth

One credit is listed at $0.001 (a tenth of a US cent). The credits you see debited are the final amount: any platform margin is already included.

What consumes credits

The LLM cost model

Each model has a price per million input and output tokens. The cost of a call is its input tokens times the input price, plus its output tokens times the output price, plus any fixed per-call cost, then the platform margin (25% of the final price) is added. Cached input tokens are billed at the model's own, cheaper cached rate. The margin does not apply to web search or other flat-priced usage.

Own-key fee

When a turn runs on your own key, your provider bills you for the tokens, and LiveContext charges a flat fee per turn based on the model's price band:

The fee is never more than what the same turn would have cost on the platform's own key, and a turn that used no tokens costs nothing. In a shared workspace the fee is billed to the workspace owner, like other usage. In Quota & Usage, own-key rows show the fee next to an estimate of what your provider charged.

The two-bucket wallet

Every subscription carries two credit buckets, which together make up your balance:

Drain order on a paid plan: the subscription bucket is used first, then the pay-as-you-go bucket. If a debit overshoots, the pay-as-you-go bucket stops at zero and the remainder is taken from the subscription bucket, so pay-as-you-go never goes negative through overshoot.

Managing your subscription

Usage and history

Settings > Quota & Usage shows your balance (split between the subscription and pay-as-you-go buckets), a usage breakdown for the last 30 days, a usage history you can filter by type, and usage analytics by day, type, provider, or model. The Workspace selector shows one workspace or All workspaces.

History shows what you took part in, as the payer or as the person who ran it. On a Team plan, a member sees their own runs in their history even though the owner was billed.

Before an expensive chat turn starts, LiveContext estimates its cost and refuses it up front if your balance cannot cover it, rather than failing halfway.

Team billing: the owner pays

In a shared workspace, usage by any member is billed to the workspace owner's wallet, not to the member. Marketplace purchases are the exception: the buyer always pays.

Refer & earn

Settings > Refer & earn holds your referral code, a share link, your progress (pending and rewarded referrals, credits earned), and a Redeem a code box for a referral or promo code you received.

• When someone who used your code subscribes, you both get 8,000 bonus credits.
• The reward is released after their first payment clears a holding period (14 days), and always lands in the pay-as-you-go bucket. If the payment is reversed, the reward is taken back from that bucket only.
• You cannot redeem your own code, each code can be redeemed once per account, and an account can use only one referral code in total.

On a self-hosted install, Refer & earn works through your connected cloud account: connect one first, and rewards land on that cloud account.

Insufficient credits and delinquency

When a check before the work starts finds that your balance cannot cover it, the action is refused with HTTP 402, so a stop for lack of credits is distinguishable from an error. The refusal is recorded in your history.

Agent credit budgets

An agent can also carry its own credit_budget, counted in the same wallet credits as everything else: before each step the agent projects what it has spent plus the likely cost of the next step, and stops when that would pass its budget. See Agents.

Self-hosted (CE): unlimited, still tracked

A self-hosted install runs in unlimited mode: every debit is recorded, the balance never runs out, and nothing blocks. Quota & Usage shows your spend in US dollars at the providers' own prices, with no platform margin. There is no Stripe and no top-up on CE.

A CE install can follow the plan of a linked cloud account. That happens only when the install is linked and at least one of its two sources is set to Cloud: the LLM source or Integration credentials, both in Settings > Cloud. An install that uses its own keys for both keeps its local plan. When the cloud plan applies, the CE pricing page mirrors the cloud subscription and sends plan changes to the cloud, and the Quota & Usage headline shows the cloud account's usage.

Any cloud plan can link an install, the Free plan included. Model calls, integration calls and web searches relayed through the cloud need a paid plan on the cloud account, and are billed to it, once per execution, so a retry never charges twice. See Self-hosting.

Troubleshooting

Related pages