Coasty Computer Use API Pricing: Every Endpoint, Every Cent
Emily Watson||6 min
Ctrl+S
Most automation tooling relies on brittle selectors or closed APIs. The Coasty computer use API lets you build agents that see the screen and act like a human. Every endpoint has a clear price, billed per operation, so you can model costs line by line. This guide lists every endpoint, the field names you need, and the exact cost in USD.
Vision endpoints
- ●POST /v1/predict costs $0.05 per call. Takes a base64 screenshot, an instruction, and cua_version. Returns actions and a status. You loop capture, predict, act until status is "done".
- ●POST /v1/sessions costs $0.10 per session. Creates a stateful trajectory memory for long-running tasks.
- ●POST /v1/sessions/{id}/predict costs $0.04 per call. Works with session memory and returns actions plus status.
- ●POST /v1/ground costs $0.03 per call. Maps a screenshot plus element description to x,y coordinates for precise clicks.
- ●POST /v1/parse is free. Converts pyautogui-style code into structured actions.
bash
curl -X POST https://coasty.ai/v1/predict \
-H "X-API-Key: $COASTY_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"screenshot": "$(base64 -w0 screenshot.png)",
"instruction": "Find the login button and click it.",
"cua_version": "v3"
}'Task runs
- ●POST /v1/runs starts a server-driven agent. Requires machine_id, task, and optional instructions (appended to the base prompt). Accepts cua_version (default "v3"; "v4" runs autonomously with a pass/fail verifier), system_prompt, max_steps, deadline_seconds, on_awaiting_human ("pause", "fail", or "cancel"), and webhook_url. Billed $0.05 per agent step.
- ●GET /v1/runs lists runs.
- ●GET /v1/runs/{id} returns the current run details.
- ●POST /v1/runs/{id}/cancel stops a run.
- ●POST /v1/runs/{id}/resume resumes a paused run.
- ●GET /v1/runs/{id}/events streams Server-Sent Events; reconnect with Last-Event-ID.
- ●States include queued, running, awaiting_human, succeeded, failed, cancelled, timed_out.
bash
curl -X POST https://coasty.ai/v1/runs \
-H "X-API-Key: $COASTY_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"machine_id": "aws-us-east-1-prod-123",
"task": "Click the submit button after the form is filled.",
"cua_version": "v3",
"max_steps": 100,
"deadline_seconds": 600,
"on_awaiting_human": "pause"
}'Workflows
- ●POST /v1/workflows defines a versioned JSON DSL of runs. Each step can be a task, assert, if, loop, parallel, human_approval, retry, succeed, or fail.
- ●Conditions are structured objects. Variables use double-brace syntax like {{inputs.x}} or stepId.field.
- ●Hard guards include budget_cents, max_iterations, and deadline_seconds.
- ●Task steps are billed $0.05 each.
bash
curl -X POST https://coasty.ai/v1/workflows \
-H "X-API-Key: $COASTY_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"name": "form_filling_workflow",
"version": "1.0",
"steps": [
{
"type": "task",
"prompt": "Fill the email field and click continue.",
"id": "fill_email"
},
{
"type": "assert",
"expression": "{{fill_email.field}}.status === "succeeded"",
"id": "email_assert"
},
{
"type": "task",
"prompt": "Submit the form.",
"id": "submit"
}
]
}'Billed $0.05 per agent step for task runs, $0.05 per vision predict call, $0.10 per session creation, $0.04 per session predict, and $0.03 per ground. All prices are in USD, 1 credit = $0.01.
Where this beats brittle automation
- ●Agents see the screen, interpret layout changes, and click exactly where a human would, instead of relying on fragile selectors.
- ●Stateful sessions maintain context across multiple calls, reducing duplicate work.
- ●Grounding lets you map a natural language description of an element to coordinates, making UI changes easier to handle.
- ●Workflows let you orchestrate complex sequences with loops, conditions, and human approval gates without writing a custom orchestrator.
- ●Transparent per-call pricing helps you forecast costs and optimize agent behavior.
Start building a reliable computer use agent with clear, line-item prices. Use the examples above as a starting point and read the full docs at https://coasty.ai/developers to get a key and explore the MCP server, scopes, and webhook signing. Visit https://coasty.ai/developers today.