> ## Documentation Index
> Fetch the complete documentation index at: https://livingcontext.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.
# Plans, pricing, and billing
> Plan tiers, seat pricing, usage credits, metering rates, spending limits, and top-ups
LivingContext pricing has three components:
1. **A base subscription** — a flat monthly price per organization, set by your plan.
2. **Seats** — paid plans add a per-seat price on top of the base subscription.
3. **Usage credits** — AI usage is metered in credits (**1 credit = \$0.01**). Every plan includes a monthly credit grant; heavier usage is covered by top-up packs on Business and Ultra.
Billing lives in the **Billing** tab of the Agents console. Only members with billing management permission (Administrators) can change plans or buy credits; most roles can view usage.
## Plan comparison
| | Free | Starter | Pro | Business | Ultra |
| ---------------------- | --------- | ------------ | ------------ | -------------- | -------------- |
| Base price / month | \$0 | \$20 | \$50 | \$200 | \$500 |
| Per seat / month | — | \$20 | \$20 | \$20 | \$20 |
| Monthly credit grant | 500 (\$5) | 2,000 (\$20) | 5,000 (\$50) | 20,000 (\$200) | 50,000 (\$500) |
| Concurrent agent runs | 2 | 3 | 5 | 10 | 20 |
| Triggers | 1 | 5 | 20 | 50 | 200 |
| Connectors | 2 | 5 | 10 | 25 | 100 |
| Side-effecting actions | — | ✓ | ✓ | ✓ | ✓ |
| Autonomous execution | — | — | ✓ | ✓ | ✓ |
| Credit top-ups | — | — | — | ✓ | ✓ |
**Side-effecting actions** are agent tool calls that change external systems — Salesforce writes, deployments, outbound calls. They are always parked behind [human approval gates](/agents/agent-platform) regardless of plan.
**Autonomous execution** covers everything that runs without a person in the loop: cron triggers, inbound webhooks, and the public run API. Interactive and assisted agent use is available on every plan.
Need higher limits, custom credit pools, single sign-on, or a self-hosted deployment? See [Enterprise](/admin/enterprise).
## How usage is metered
Every model call an agent makes is metered at the moment it completes and converted to credits:
```text theme={null}
cost = (input tokens × input rate + output tokens × output rate) × 1.2
credits = cost ÷ $0.01
```
* Rates are per million tokens and vary by model (table below).
* The **1.2× platform multiplier** is applied to the raw model cost.
* Costs are recorded per message and per run, so the Billing tab can attribute spend to specific agents and runs.
### Billed rates by model
Credits charged per **1 million tokens** (multiplier included):
| Model | Input | Output |
| ----------------- | ----- | ------ |
| Claude Opus 4.5 | 600 | 3,000 |
| Claude Sonnet 4.6 | 360 | 1,800 |
| Claude Haiku 4.5 | 120 | 600 |
| Gemini 3.1 Pro | 240 | 1,440 |
| Gemini 3.5 Flash | 60 | 360 |
| Gemini 2.5 Pro | 150 | 1,200 |
| Gemini 2.5 Flash | 36 | 300 |
Models not in the catalog bill at a conservative premium rate (600 input / 3,000 output per 1M tokens) so an uncatalogued model can never undercharge. The same model bills identically no matter which provider route serves it.
**Worked example:** one agent turn on Claude Sonnet 4.6 that reads 10,000 tokens and writes 2,000 tokens costs (10,000 × $3 + 2,000 × $15) ÷ 1M × 1.2 = \$0.072 → **7.2 credits**. The Free plan's monthly grant covers roughly 70 such turns.
## When credits run out
When your credit balance reaches zero, **agent runs are blocked** with a clear message ("Out of credits — top up or upgrade"). Blocking applies to every agent surface: chat, triggers, webhooks, and the API.
Everything that doesn't consume AI usage keeps working — reading docs, search, the Inbox for already-generated proposals, change request editing, merges, and integrations syncing.
Balance replenishes automatically with the next monthly grant, immediately on plan change, or instantly with a top-up.
## Concurrency limits
Each plan caps how many agent runs can be active at once (queued, running, or awaiting approval). Starting a run beyond the cap fails with an actionable message that names your current usage and plan — wait for a run to finish, or upgrade.
## Top-ups
Business and Ultra plans can buy additional credits at any time:
* **Card checkout** — a hosted Stripe checkout, from 100 up to 500,000 credits per purchase. Credits land on your balance as soon as payment completes.
* **Manual grants** — for deployments without card payments configured (for example, self-hosted or invoiced accounts), credits are granted directly by agreement, up to 50,000 per grant.
Purchased credits are priced at the same rate as granted credits: 1 credit = $0.01, so a 1,000-credit pack is $10.
On Free, Starter, and Pro, top-ups are not available — if you regularly exhaust your grant, upgrading is the intended path, and it's usually cheaper than the equivalent top-up volume.
## Plan changes and seats
* Plan changes take effect immediately: new limits apply at once, and the new plan's monthly credit grant is applied on the change.
* Seats are billed at \$20/month each on all paid tiers. Your seat count is managed alongside the plan in the Billing tab.
* Downgrading never deletes anything — if you exceed the lower plan's trigger or connector caps, existing items stay but new ones can't be created until you're back under the cap.
## Auditability
Every credit movement — grants, usage debits, top-ups, adjustments — is recorded in an append-only ledger with a running balance, visible in the Billing tab. Usage records carry the model, token counts, and originating run, so finance and engineering see the same numbers.
## Frequently asked questions
Credits meter AI usage: agent conversations, document generation, suggestions, and AI editor actions. Browsing, searching, editing by hand, and merging change requests are free.
Spending is naturally capped by your credit balance — LivingContext blocks agent usage at zero rather than billing overage. On plans without top-ups, your monthly grant is a hard ceiling.
The monthly grant sets your balance for the cycle; topped-up credits remain until spent.
Yes — organizations can bring their own Anthropic API key in **Settings → AI credentials**, which routes usage to your own provider account. See [Enterprise](/admin/enterprise).
Administrators. Other roles can read billing status. See [Roles and permissions](/admin/roles-and-permissions).