Using the API with your plan
Call Dropstone over HTTP with an API key. The endpoint is OpenAI-compatible, billed pay-per-use from your credit balance, and works from any language. Keys, errors, rate limits, and what the API does not carry.
The Dropstone API is an OpenAI-compatible endpoint for code that should not depend on a signed-in surface: a CI job, a serverless function, a service in another language. Create a key, point a client at Dropstone, and you are calling the same three models as Chat and the CLI.
The endpoint reference is at docs.dropstone.io/cli/api. This page covers keys, billing, and what to expect.
Create an API key
- Open Settings in your dashboard.
- Under API keys, choose Create key and give it a name.
- Copy the key. It looks like
dsk_live_...and it is shown once.
Treat the key like a password: never commit it, and never ship it in a browser bundle. Anyone holding it can spend the balance behind it. You can see when each key was last used and revoke any of them from the same page.
Billing: pay per use
API-key requests are billed pay-per-use from your credit balance. Plan allowances do not apply to them.The weekly allowance belongs to the signed-in surfaces: Chat, the CLI, and a local SDK session. The API is metered separately, so a pipeline's traffic never eats the allowance you work in, and a busy week in Chat never stops the pipeline. Top up from Billing in the dashboard.
A key whose balance is empty is refused with an error saying so and a link to top up, rather than being served and charged later. There is no free tier on the API.
OpenAI-compatible
The endpoint follows the OpenAI chat completions format, so most existing clients work by changing the base URL and the key:
import OpenAI from "openai"
const openai = new OpenAI({
baseURL: "https://api.dropstone.io/api/v1",
apiKey: process.env.DROPSTONE_API_KEY,
})
The models are dropstone-fast, dropstone-pro, and dropstone-heavy; see Choose a model. Requests work from any language that can make an HTTP call.
What the API does not carry
The endpoint is stateless, and deliberately so:
- No memory. It does not read your account memory, and nothing it answers is recorded there. The same input gives the same answer, whatever you taught Dropstone last week.
- No tools and no agent loop. It answers a chat completion. Reading a codebase, editing files, and running commands belong to the SDK local client and the CLI.
If you want your code to use the agent and its memory, use the SDK's local client. See One memory across every surface.
Rate limits
Requests are rate-limited per IP, well above ordinary use, to protect the service. A burst that trips the limit is refused, so back off and retry rather than tightening the loop.
Where requests go
API requests are processed in the United States, the same as every other surface. See Where your requests run.
Organizations
Members of an organization create keys under their own account, and the balance behind a key is the key owner's. Organization admins can see API usage by member on the organization dashboard, and a key stops working when the account behind it is deactivated, which is what happens when a member is removed through SCIM.
Related articles
- Dropstone SDKA typed TypeScript and JavaScript client for the Dropstone agent, with two entry points: one for headless API calls from CI or a serverless function, one that runs the agent locally with the same account memory as the CLI and Chat.
- Install and run your first sessionInstall the Dropstone SDK, make a headless API call with an API key, and run a local agent session that carries your account memory. Includes streaming, structured output, and using the OpenAI SDK against Dropstone.
- One memory across every surfaceThe account memory that follows you across Dropstone Chat, the CLI, and the SDK, what shares it and what does not, and how to keep an integration's learning out of your own memory.
- Choose a modelDropstone offers three models, Fast, Pro, and Heavy, tuned for different kinds of work. Fast and Pro are available on every plan. Heavy is included with paid plans. Here is how to pick.
- Where your requests runDropstone's servers and the inference providers that run its models are hosted in the United States. Requests are never routed through providers outside the US. What that means for your data.