---
title: "Billing"
description: "Fund managed agent work with prepaid credits and inspect usage by model and task."
canonical_url: "https://hoplite.sh/docs/workspace/billing"
markdown_url: "https://hoplite.sh/docs/workspace/billing.md"
---

# Billing
URL: /docs/workspace/billing
LLM index: /llms.txt
Description: Fund managed agent work with prepaid credits and inspect usage by model and task.
Related: /docs/threads/cost, /docs/agent/models, /docs/workspace

# Billing

Hoplite bills on **prepaid credits**. Your workspace holds a credit balance; agent runs draw it down based on the tokens they actually use, at per-model rates.

## Free plan

Free workspaces receive a **monthly allowance of included credits** and can have up to **five projects**. Only part of the included allowance can be used per UTC day; purchased prepaid credits remain available after that daily ceiling is reached. Billing displays included usage as a percentage on every plan, including trials, rather than disclosing its dollar size. Daily limits still apply and refresh at the displayed reset time; purchased balances and credit spend remain dollar-denominated. A project can contain multiple repositories, so the limit counts projects rather than repositories attached to them.

## Credits

One credit represents US$1 of billable usage.

The billing page under **Settings → Workspace → Billing** shows:

| Balance | Meaning |
| --- | --- |
| **Available** | What you can spend right now |
| **Held** | Credits reserved by in-flight runs, released back (minus actual usage) when they finish |
| **Granted / used** | Lifetime totals, including any starter grant |

## Top-ups

Buy credits through Stripe checkout from the billing page; a Stripe billing portal handles receipts and payment methods. If a run stalls on an empty balance, an **Add credits** dialog appears right in the thread so you can top up and retry without losing your place. You can also [commit and push the run's uncommitted edits](/docs/threads/run#save-work-after-an-out-of-credits-stop) before topping up.

A top-up never restarts stopped threads on its own. Once runs can start again, owners and admins see how many threads stopped when credits ran out in the last 14 days, with **Retry all** on the billing page. After you confirm, each thread retries the same way as its own retry button, so the usual credit and access checks still apply. If credits run out again partway through, or the workspace spend limit or billing refuses a retry, the remaining threads stay stopped. Other members' private threads and threads that run on a local agent host are not included, and a thread started by a Linear delegation appears only after that delegation's own retries have finished. The offer is hidden while a workspace spend limit has no room left for another run.

## Free signup and Team trials

New workspaces can finish signup on Free without choosing a paid plan. On hosted
Hoplite, if the Free-plan payment-method gate is enabled, enrolling a new
workspace opens a Stripe setup checkout for a billing address and card. This
saves the payment method without charging it. Prepaid cards are not accepted;
this is separate from Hoplite's **prepaid credits**, the workspace balance used
for agent runs. Existing Free workspaces and workspaces downgrading from a paid
plan do not repeat the enrollment gate, and local or self-hosted deployments do
not use the hosted signup flow.

Team is currently advertised without a free trial; new workspaces start on
Free and can upgrade to Plus or Team. Existing Team trials continue until they
end. If Billing presents an active trial offer to an eligible Free workspace,
it supports up to three seats and shows the trial length and purchase terms at
checkout. The workspace must have no subscription or previous trial history.

A trial's included credits are granted for one seat whatever its seat count,
and are subject to a per-UTC-day ceiling; purchased prepaid credits are never
capped. Each card can start one trial: a trial started with a card that already
ran one is cancelled.

<Callout type="info" title="Current purchase options">
Hosted Hoplite sells **Plus** (per seat, charged up front, no trial) and
**Team** (per seat) to new purchasers. Plus has Team's features with lower
limits: fewer concurrent sandboxes, new threads per hour and active
automations, 5-minute instead of 1-minute schedules, and sandboxes up to
Large instead of X Large. Plus workspaces from before November 2026 keep
X Large. Prices and included allowances also differ. Each plan includes a monthly usage allowance
per seat; see the pricing page for purchase options and billing settings for
usage. Existing Pro subscriptions keep the price
and per-seat allowance they signed up under until they choose a different plan
themselves; existing Scale subscriptions remain supported, but Scale is no
longer offered for new purchases.
</Callout>

## Plans, seats, and subscription changes

Owners and admins preview plan, billing-interval, and paid-seat changes before
applying them. Changes that take effect immediately — including a trial
conversion — open a hosted Stripe confirmation page; ending a trial this way
ends it immediately, and any amount shown is due now. Return to Hoplite after
Stripe confirms the update so Billing can refresh the authoritative state.

No-charge downgrades can be scheduled for renewal. Billing shows the pending
change and lets you cancel it to keep the current plan, interval, and seat count.

**Downgrade to Free** or **Continue on Free** checks the workspace against Free's
one-seat and five-project limits. The dialog lets you remove extra members,
cancel invitations, and delete excess projects before switching. Deleting a
project permanently removes its threads, automations, and environment variables;
connected GitHub repositories are unchanged. An active subscription ends at
renewal, while an expired workspace moves to Free immediately. Cancelling a
past-due or unpaid subscription ends it immediately and stops renewal payment
retries rather than waiting for the next billing date. Prepaid credits remain
available.

**Change seats** opens a seat-only dialog that keeps the current plan and
billing interval. The seat count can't go below the workspace's members plus
pending invitations, and the dialog shows the new total before you confirm.
While a scheduled change is pending, cancel it before adding seats; removing
seats replaces it.
A pending renewal change must be cancelled before a different immediate change
can be confirmed.

An unpaid renewal does not grant the new period's included allowance until its
payment succeeds.

## After access expires

Projects, threads, history, settings, and billing remain readable after subscription access expires or the seat limit is exceeded. Owners can restore access from Billing.

New cloud runs and resumptions are blocked, and active cloud runs are cancelled after access ends. Automation triggers, review-loop passes, and preview starts or resumes are also blocked. An already-active preview is not proactively stopped solely by the billing transition. Explicit local-agent threads remain usable.

## What's metered

Each model call records input, output, cached, and reasoning tokens with its provider, model, run, and cost. Reasoning tokens are billed as output. [Model routing](/docs/agent/models) determines whether a call uses workspace credits, a connected subscription, or your provider key.

Hoplite pays for the [Slack](/docs/integrations/slack) and iMessage assistant's own conversation, so it draws no workspace credits and never uses a member's subscription or your provider keys. It has a daily limit per person and per workspace; when one is reached, the assistant replies with when it resets. Coding threads the assistant starts or messages are billed normally.

Sandbox compute draws workspace credits. Daytona records can arrive up to 48 hours late, so a deleted workspace may retain **Pending** VM cost while charges settle. Modal sandboxes are charged a fixed hourly price for their size, billed per second only while the agent is working: from the moment each run starts until five minutes after it ends. Idle time between runs, and a sandbox left running after a preview or terminal, is free. Charges settle while runs are in progress. A sandbox's size is its box: it bursts up to that CPU and memory as the work needs it. Sandboxes without a configured size are Medium, clamped to your plan (Free tops out at Small, Plus at Large). An unsized Docker Compose VM runs at 4 vCPU / 8 GiB and is billed as Medium. A custom size bills as the smallest preset that covers both its CPU and memory; anything larger than X Large bills as X Large.

| Size | Resources | Price per hour |
| --- | --- | --- |
| X Small | 1 vCPU · 4 GiB | $0.08 |
| Small | 2 vCPU · 8 GiB | $0.16 |
| Medium | 4 vCPU · 16 GiB | $0.30 |
| Large | 8 vCPU · 32 GiB | $0.75 |
| X Large | 16 vCPU · 64 GiB | $1.50 |

Modal sandboxes started before this billing shipped in October 2026 were not charged and show no VM cost.

## ChatGPT subscription models

Calls routed through your connected ChatGPT subscription use that plan and incur zero Hoplite workspace-credit cost. Selecting an OpenAI model alone does not guarantee that route: eligible calls can use platform billing or BYOK instead. Check the payer in the thread's cost breakdown or Usage, and see [model routing](/docs/agent/models#automatic-chatgpt-api-routing) for eligibility and fallback behavior.

Automatic context compaction normally appears in usage attribution with zero workspace-credit cost because Hoplite runs that summarization on a platform-funded model. If platform summarizer resolution fails, compaction falls back to the thread's session model and follows that model's normal workspace-credit, BYOK, or ChatGPT-subscription attribution.

## Usage

**Settings → Workspace → Usage** charts credit spend by model per day or week and breaks it down by member. Owners can also break it down by project and open a member's projects and threads, and **Export CSV** downloads the selected period. Usage tracking is available on Plus, Team, legacy Pro, and Scale.

On hosted Hoplite, dollar totals on Usage and Billing show settled spend from
prepaid credits, excluding charges covered by the plan's included allowance.
Calls and token counts still include that activity, so they can be nonzero when
credit spend is zero. **Included usage** shows the percentage of the billing
period's allowance used, not a dollar charge. Included credits are used before
prepaid credits.

Owners and admins can also configure **Settings → Workspace → Billing → Usage limit**. When enabled, this caps credit spend in the UTC calendar month; usage covered by the plan's included allowance does not count toward the limit. Hoplite stops a new run before its credit hold would push the workspace over the limit. Disable the policy to remove the cap, or raise it and send the message again; the usage limit does not add credits to the workspace balance.

## Cost per thread

Spend is also visible where the work happens. Each thread's context-gauge card shows running cost, attributes usage to individual runs, and marks runs billed to your own provider keys with a BYOK badge — see [Track cost](/docs/threads/cost). Compatible Anthropic and OpenAI provider keys are configured under **Settings → Workspace → Models** when BYOK is enabled.
