Pay per builder, not per teammate.

Add your whole team for free. You pay only for builders who ship changes.

Free
Try Gently on a real project.
$0
just you
  • 1 builder
  • 1 concurrent thread
  • $5 usage to start
  • Bring your own AI
Request access
Starter
Solo builders and two-person teams.
$25 / builder
billed monthly, up to 2 builders
  • Unlimited teammates, only builders pay
  • 2 concurrent threads per builder
  • $12.50 of usage included per builder each month
  • Agent workspaces included
Request access
Team
Teams shipping every day.
$49 / builder
plus a $99 team fee each month
  • Example: 6 builders cost $393 / month
  • Unlimited teammates, only builders pay
  • 2 concurrent threads per builder
  • $25.00 of usage included per builder each month
  • Deep review and automations
Request access
Enterprise
For organizations that need security and support.
Custom
annual
  • Your own cloud or a dedicated node
  • SSO, audit, and retention controls
  • Volume pricing and an SLA
Talk to us

2 extra concurrent threads: $49 a month on any paid plan · Annual billing: 20% off

Standard plans have a 0% bring-your-own AI platform rate.

Connect your own OpenAI or Claude account, API key, or local model. On our standard plans, you pay the AI provider and Gently adds no platform charge for those model calls. Custom enterprise agreements may define a different bring-your-own AI platform rate. If Gently provides the model or service, we charge the provider price plus 5%. Included usage covers Gently-provided AI and live cloud app runtime. Purchased credits roll over. Included usage resets each month.

Agent workspaces are included. AI and live apps use credits.

Builder seats include the Gently platform and agent workspace time. Gently-provided AI and live cloud app runtime draw from your shared usage balance. Idle sandboxes sleep and cost nothing. Gently Desktop does not charge for runtime on your own Mac.

A thread is one job, but it can make several AI calls.

Sub-agents, code review, and browser checks inside a job do not use more concurrent threads. The AI calls they make can still count toward usage. Your plan sets how many top-level jobs can run at once.

A builder is anyone who ships changes.

Admins assign builder seats to people who change code. An admin only needs a seat if they also change code. Everyone else can ask questions, research, plan, and open previews without a builder seat; any Gently-provided AI or cloud runtime still draws from the workspace usage balance.

Worked example

Suppose all six people are builders. In one month, they complete the work below. Their 2 vCPU, 4 GB cloud sandboxes run apps and services for 563.5 aggregate hours across those threads. This invoice shows the cost and the Team plan's included usage.

74pull requests merged across 3 repositories
57completed delivery workstreams
821questions answered
132research requests completed
How the sandbox rate is calculated
2 vCPU × $0.05 / vCPU-hour$0.10 / hr
4 GB × $0.015 / GB-hour$0.06 / hr
2 vCPU, 4 GB sandbox$0.16 / app-hour

We charge by the second while an app or service runs. We add the running time from concurrent sandboxes.

Example monthly invoice
Team · 6 builders · monthly billing
USD
Team base price
$99.00
1 × $99.00
Builder seats
$294.00
6 builders × $49.00 each month. Other teammates are free.
AI with your own account
$0.00
On Team’s standard 0% bring-your-own AI platform rate, Gently adds no charge for model calls covered by your connected account.
Gently-provided AI and tools
$42.50
Main and supporting AI calls supplied by Gently, at provider price plus 5%.
Live cloud app runtime
$90.16
563.5 app-hours × $0.16 / hour
Team usage included
−$132.66
6 builders × $25.00 = $150.00 included each month
Usage overage
$0.00
You pay only for usage above the monthly included amount.
Total due$17.34 of this month's included usage remains
$393.00
With annual billing, the platform subscription is $3,772.80 per year ($314.40 monthly equivalent). The $150.00 usage allowance still resets each month.
Why do I pay for both builders and usage?

They pay for different things. A builder seat pays for access to build and ship with Gently, including the platform, agent workspaces, and concurrent-thread capacity. Usage covers variable costs created while work runs: Gently-provided AI and live cloud app runtime. Every paid builder adds included usage to one shared workspace balance, and we use that balance before any purchased credits or allowed overage.

What exactly counts as AI usage?

When Gently supplies a model or AI service, each provider-backed call counts toward usage—not just the final answer. On standard plans, calls covered by your own connected account have a 0% Gently platform rate; custom enterprise agreements may define a different rate. The separate answer below explains which supporting services may still count.

  • The main AI work for questions, research, planning, building, and automations.
  • Supporting work such as routing requests, checking completed responses, summarizing long conversations, running sub-agents, searching the web, making extra model calls inside tools, understanding screenshots and images, generating images, providing advisor guidance, and narrating browser recordings.
  • Delivery stages such as plan review, code review and specialist review, applying or verifying findings, final and goal verification, and AI-assisted commit or pull-request preparation.
  • Other AI tools such as prompt coaching, Ask the Database, voice transcription and cleanup, project setup assistance, and AI connection tests.
  • Provider retries and fallback calls when more than one model call is needed to complete the work.

Text models are priced from input, output, and cached tokens. Search, images, transcription, and narration use the provider’s relevant unit, such as requests, images, audio duration, or spoken characters.

Do all AI calls inside one thread count separately?

Yes. One thread uses one concurrent-thread slot, but it can involve several AI calls. For example, a build may use the main model, sub-agents, summaries, searches, and review stages. Each metered call contributes to usage even though the whole job still occupies only one top-level thread slot.

Do you charge for AI when I use my own account or API key?

On standard plans, Gently’s platform rate is 0% for model calls covered by your own API key, local model, or connected ChatGPT or Claude subscription; you pay that provider directly, if applicable. Custom enterprise agreements may define a different bring-your-own AI platform rate. Supporting services that are not covered by your connection—such as separate classification, summarization, search, vision, image, voice, narration, or advisor calls—can still use Gently-provided services and count toward usage. When Gently provides the model or service, we charge the provider price plus 5%.

Who counts as a builder?

A builder is anyone an admin allows to change code. Admins only need builder seats when they also change code. Everyone else can ask questions, research, plan, and review previews without a builder seat. Their Gently-provided AI and cloud runtime still draw from the workspace usage balance.

What happens when I run out of usage?

You can pause until next month, use purchased credits, or set a monthly limit for extra usage. We alert admins at 50, 80, and 100%.

Can I switch plans?

Yes. You can switch plans at any time. We prorate plan and builder changes on your subscription.