Pricing
What Evergreen charges for, what a typical session costs, and what's included in your first week.
Evergreen charges for what your sessions use and nothing else. There are no seat fees, so inviting the whole team costs nothing until they start sessions. Usage is paid from a balance you add funds to.
What you pay for
You pay Evergreen for the tokens the agent uses, the time it spends working (like CI minutes), the disk its workspace keeps, and the embeddings and reranking behind search and knowledge. The models cost exactly what the provider charges Evergreen, discounts included and never marked up.
That covers your sessions and the model calls Evergreen makes for your workspace, such as titles, summaries, search intent, and reviews, which show as Background work. On your own key, your provider bills your sessions’ models instead. See Bring your own key.
The platform fee is the same on every model and on your own key too. It covers what Evergreen adds around the model, such as the review before every pull request, summaries, workflows, and the shepherd. Agent time is the same whichever model the agent runs. Search and knowledge run on Amazon Bedrock at AWS’s own price, on your own key too.
Each session shows its own charges, and background work and search are listed separately.
The rate card
| What you use | Price | Notes |
|---|---|---|
| Platform fee | 15¢ per million tokens | The same on every model, and on top of the model’s own price. |
| Agent time | $1 per hour | Counted while the agent works. Waiting and sleeping between turns are free. |
| Workspace storage | 35¢ per GB-month | Counted while a session’s workspace is kept. |
| Search and knowledge, embeddings | 12¢ per million tokens | For your workspace’s search and knowledge. |
| Search and knowledge, reranking | $2 per thousand searches | For your workspace’s search and knowledge. |
A typical session
A typical session uses about 8M tokens, 10 minutes of agent time, and a 17 GB workspace kept for 1.9 days. Evergreen’s part of that is $1.74, plus the model at the provider’s price. On Claude Opus 5.5, the model costs about $4.27 at the provider’s list price, so a typical session comes to about $6 in all.
| Line | What it’s based on | Cost |
|---|---|---|
| Platform fee | 8M tokens | $1.20 |
| Agent time | 10 minutes of work | 16¢ |
| Workspace storage | 17 GB for 1.9 days | 38¢ |
| Evergreen’s part | $1.74 | |
| Model | Claude Opus 5.5, at the provider’s list price | About $4.27 |
| In all | About $6 |
Your sessions will differ. A long session on a big task uses more tokens and more agent time, and a quick one uses less. The session page shows what each session has cost. A session’s cost includes the model as well as Evergreen’s part.
Note
Prices are in US dollars and are the same for every workspace unless Evergreen agreed a different plan with you. Charges already booked keep the price they were booked at.
Your first week
A workspace you own pays no platform fee, agent time, workspace storage, or search for seven days, counted from its first session. On your own Anthropic, OpenAI, or xAI key you don’t need a card, and the week starts on its own when you run your first session. Models and knowledge are still billed from the first turn: models at the provider’s price (or by your provider, on your own key), and knowledge at AWS’s price. On Evergreen’s models, that means you’ll add a card and funds before your first session. Each person gets one free week in the first workspace they own that runs a session.Workspace settings › Billing shows what your free week is waiting for, such as a card or your first session. Once it starts, the page shows when it ends and what model inference has cost so far. You also get an email when the week starts, two days before it ends, and once it has ended.
When usage is booked
Usage is booked as it happens, in the Pacific time zone’s days. Your balance, the statements, and the daily limits on auto-reload all use the same Pacific billing day.
Last updated