Essentials
Plans, credits, and pricing
Plan prices and allowances, what uses credits, how AI Gateway calls are priced, limits, and which plans unlock each feature.
Credits are the one currency for coding, planning, verification, browser tasks, and AI calls from your apps. One credit is worth $0.10.
Check your balance with /balance in the workspace or CLI.
Plans
| Plan | Monthly | Annual | Credits | Adds |
|---|---|---|---|---|
| Free | $0 | $0 | 20 a month | Coding agent in the browser and terminal, browser agents, publishing for 7 days |
| Plus | $10 | $99 | 100 a month | More monthly credits |
| Premium | $49 | $499 | 500 a month | Model selection, published apps that stay live, custom domains |
| Pro | $199 | $1,999 | 2,500 a month | More monthly credits |
| Team | $499 | $4,990 | 5,000 a month, pooled | Unlimited team seats |
| Enterprise | Custom | Custom | Per agreement | A dedicated solutions engineer |
Each plan includes everything in the plans above it. A team workspace without a plan gets its 20 free credits once, not every month. See pricing for current offers, and manage your plan, top-ups, and auto-recharge in Billing.
What uses credits
| Work | How it is charged |
|---|---|
| Coding and planning | Measured model work; model choice and effort affect usage |
| Verification | Each judge call is charged like a turn |
| Images, video, and audio the agent generates | Charged per generation |
| Fast and Deep specs | Variable; see the Architect overview for estimates |
| Browser tasks | A floor per task at submission, and more for successful tasks that do more work |
| Browser Agents Local | A small fraction of a credit per plain-words action; every other Local tool is free |
| AI calls from your apps | Each call at the model’s credit price; see AI Gateway calls |
Every browser task is charged a 0.1-credit floor when you submit it, so you need available credits to start. When the task finishes, it is charged any usage above the floor, whether it succeeded or not. Only an ERROR or EXHAUSTED task keeps the floor and nothing more.
A browser submission can be rejected because of an account limit even when the overall balance is positive. Browser trial limits and in-flight caps are separate from your credit allowance.
AI Gateway calls
Calls to the AI Gateway, including the AI features in apps you build on pre.dev, are charged from each call’s metered usage at the model’s credit price. The price per model is listed on GET /v1/models: each row carries a predev object with credits_per_m_input and credits_per_m_output and, where the model has them, prices such as credits_per_m_cache_read, credits_per_m_cache_write, credits_per_image, and credits_per_request. Read those numbers rather than deriving prices yourself.
Worked example: a model whose row lists credits_per_m_input: 20 and credits_per_m_output: 100 costs 0.2 credits for 10,000 input tokens plus 0.2 credits for 2,000 output tokens, so a call of that size settles at about 0.4 credits.
- Plugin costs, such as web search, are part of the call’s charge.
- With
modelsfallbacks, you pay the price of the model that answered. - Streams settle when they end. If your client disconnects mid-stream, the call is charged for what the model generated.
- Video jobs are charged when a status poll reports completion, and submitting one needs at least 10 credits available. See images and video.
- Catalog reads are free.
Non-streaming responses report the charge in x-predev-credits-charged and the balance in x-predev-credits-remaining. GET /v1/usage breaks the last 1 to 90 days down by model and path.
Free allowance. A Free workspace can spend 5 credits on gateway calls in total, whatever its credit balance. After that, every call returns 402 with code subscription_required until the workspace subscribes.
Out of credits. A workspace without enough credits for a call gets 402 with code insufficient_credits before anything is sent to the model, so nothing is generated or charged. Turn on auto-recharge in billing to keep a live app running when the balance would otherwise reach zero. See AI Gateway errors.
Limits
| Limit | Value |
|---|---|
| Browser tasks per request | 1,000 |
Browser tasks running at once within a run (concurrency) | 1 to 20, default 5 |
| Browser tasks in flight per account, queued or running | Set by your plan; read cap from GET /browser-agent-status |
| AI Gateway requests per workspace | 600 a minute; 30 a minute on the Free plan. Catalog reads do not count. |
Check usage and recover
/balanceshows the account balance.- The credits pill next to the composer shows what the current project has used. Select it for Credits on this project: a newest-first ledger of turns with the phase, model, tokens, and credits for each.
/usagein the CLI shows the same breakdown in the terminal.- In the CLI,
/topupand/upgradeopen billing.
A run paused for insufficient credits can pick up where it stopped after you add credits; check its current status before launching duplicate work.
For integrations, GET /credits-balance returns the effective balance, and GET /browser-agent-status reports the current browser in-flight cap.
API and MCP access
The Architect API and the MCP server are included with the Premium, Pro, Team, and Enterprise plans. Plus accounts, and team workspaces without an active plan, can connect and generate 3 specifications as a trial; after that, generation returns 403. Free accounts cannot authenticate to the Architect API or MCP.
Browser Agents REST and the AI Gateway accept every account, including Free. See authentication for how each interface checks keys and errors and retries for access errors.