Guide

Coasty Computer Use API Pricing, Every Endpoint, Every Cent

Name: Coasty AI Employee
Brand: Coasty
Price: 19 USD
Availability: InStock
Rating: 4.8 (1250 reviews)

David Park|July 5, 2026|7 min

Del

You want to build a computer use agent that drives browsers and desktops like a human, not a brittle selector library. You need to know the cost before you ship. The Coasty computer use API bills by step, by call, and by session. This guide lists every endpoint, every field, and every cent so you can plan your budget and integrate with confidence.

Billing model at a glance

Coasty uses a prepaid USD wallet. One credit equals $0.01. You pay per agent step, per prediction, and per session call. No hidden fees. No surprise monthly bills. You control spend with scopes, idempotency, and webhooks.

●Prepaid USD wallet. 1 credit = $0.01.
●Billed $0.05 per agent step for Task Runs.
●Vision endpoints: $0.05 for /v1/predict, $0.04 for /v1/sessions/{id}/predict, $0.03 for /v1/ground.
●POST /v1/parse is free.
●Webhooks are signed with HMAC for security.

Vision: screen understanding

The API needs to see what is on screen to act like a human. The vision endpoints take a base64 screenshot and text instructions. They return coordinates and actions. You loop capture, predict, act until the status is done.

●POST /v1/predict costs $0.05. Request body includes base64 screenshot, instruction, and cua_version. Response includes actions and status.
●POST /v1/sessions costs $0.10 to create a stateful trajectory memory. Then POST /v1/sessions/{id}/predict costs $0.04 per step.
●POST /v1/ground costs $0.03. It maps a screenshot + element description to x,y coordinates.
●POST /v1/parse is free. It turns pyautogui code into structured actions.

bash

curl -X POST https://coasty.ai/v1/predict \
  -H "Authorization: Bearer $COASTY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "screenshot": "iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAADUlEQVR42mNk+M9QDwADhgGAWjR9awAAAABJRU5ErkJggg==",
    "instruction": "Click the Save button.",
    "cua_version": "v3"
  }'

Task Runs: autonomous agents

The server drives an agent to completion. You provide a machine_id, a task, and options. The API bills $0.05 per agent step. You can pause, fail, cancel, or resume a run. You can stream events with Server-Sent Events.

●POST /v1/runs starts a run. Fields: machine_id, task, cua_version (default v3, v4 adds a pass/fail verifier), instructions (appended to the base prompt), system_prompt, max_steps, deadline_seconds, on_awaiting_human (pause, fail, cancel), webhook_url.
●GET /v1/runs lists your runs.
●GET /v1/runs/{id} returns the run status.
●POST /v1/runs/{id}/cancel stops a run.
●POST /v1/runs/{id}/resume continues a paused run.
●GET /v1/runs/{id}/events streams Server-Sent Events. Reconnect with Last-Event-ID header.
●States: queued, running, awaiting_human, succeeded, failed, cancelled, timed_out.

python

import os
import requests

def start_run():
    url = "https://coasty.ai/v1/runs"
    api_key = os.getenv("COASTY_API_KEY")
    resp = requests.post(
        url,
        headers={"Authorization": f"Bearer {api_key}"},
        json={
            "machine_id": "vm-prod-001",
            "task": "Open Chrome and go to https://example.com",
            "cua_version": "v4",
            "max_steps": 50,
            "on_awaiting_human": "pause"
        }
    )
    resp.raise_for_status()
    run = resp.json()
    print("Run ID:", run["id"])
    return run

start_run()

Workflows: versioned DSL of runs

Workflows let you orchestrate runs with a versioned JSON DSL. You can nest tasks, asserts, loops, and conditions. Each task step is billed $0.05. Variables like {{inputs.x}} or stepId.field reference inputs and step outputs.

●POST /v1/workflows creates a workflow.
●POST /v1/workflows/{id}/runs starts a workflow run.
●POST /v1/workflows/runs starts an ad-hoc inline workflow.
●Step types: task, assert, if, loop, parallel, human_approval, retry, succeed, fail.
●Conditions are structured objects. Use budget_cents, max_iterations, deadline_seconds as hard guards.

Machines: provision real desktops

POST /v1/machines provisions a cloud VM you can start, stop, and snapshot. The agent drives real desktops, browsers, and terminals, not just API calls.

●POST /v1/machines creates a machine.
●You can start, stop, and snapshot machines from the API.
●Use machine_id in Task Runs to target a specific machine.

1 credit = $0.01. Pay per agent step, per prediction, and per session call. No hidden fees.

Where this beats brittle automation

Traditional automation relies on brittle selectors, XPath, and API contracts that change. A computer use agent sees the screen, understands natural language, and acts like a human. It can handle UI updates, missing elements, and dynamic layouts without breaking. With the Coasty computer use API you get a single, predictable billing model and real, verified performance (85.6% on OSWorld using our in-house model).

●Sees the screen and acts like a human.
●Works with browsers, desktops, and terminals.
●Handles UI changes without breaking.
●Predictable per-step pricing.
●Verified benchmark results on OSWorld.

You now know every endpoint, every field, and every cent for the Coasty computer use API. Build a browser automation agent, a desktop pilot, or a QA test runner. Start with a machine, launch a run, and scale with workflows. Get your API key at https://coasty.ai/developers to build your first computer use agent today.