Skyvern vs Ui.Vision

Skyvern

• Operations teams collecting invoices or reports from vendor portals that have no API
• Developers replacing selector-based Selenium or Playwright scripts
• Teams that need to self-host browser automation

Ui.Vision

• Testers and developers familiar with Selenium IDE-style commands
• Automating sites or desktop apps where selectors fail but images and text can be seen
• Teams that need automation to run entirely on their own machines

Skyvern

• The cloud Free plan is 5,000 credits, about 170 actions, and the pricing page does not show it renewing each month.
• 2FA and TOTP support, team workspaces, residential proxies and state or city targeting start on Pro at $149 a month.
• Custom code blocks, human-in-the-loop steps and HIPAA compliance are listed only under Enterprise.
• The self-hosted edition is AGPL-3.0 and needs your own LLM provider; proxies, anti-bot measures and CAPTCHA solving are described as part of Skyvern Cloud.

Ui.Vision

• Desktop automation and local file access need the separate XModules desktop app; the extension alone automates only the browser.
• Browser Vision input requires Chrome or Edge, so it is not available in the Firefox extension.
• The PRO and Enterprise editions add support and update management, not extra automation features.
• The hosted AI assistant sends chat content, page data and screenshots to the provider for processing, and AI PRO has no uptime guarantee.

Skyvern

Vision-based browser agents

Agents combine computer vision with LLM reasoning to read a page, so a workflow keeps working after a site changes its layout. Each run has a summary of the actions taken and why.

Several ways to build a workflow

Describe the task to a copilot in plain English, upload a written procedure, record yourself doing it in a browser, or assemble blocks in a visual editor.

Logins, 2FA and CAPTCHAs

A credential vault stores passwords and 2FA codes without exposing them to the LLM, and connects to 1Password and Bitwarden. Runs can pass TOTP prompts and solve CAPTCHAs.

Structured data extraction

Data pulled from a site is returned in a schema you define, as JSON or CSV, and documents can be downloaded from portals that sit behind a login.

Proxies and geo-targeting

Cloud runs go through datacenter or residential proxies, with targeting by country on the lower plans and by state or city on Pro and Enterprise.

API, SDKs and MCP

Workflows can be triggered through a REST API, Python and TypeScript SDKs or webhooks, and Skyvern can act as a tool for MCP-compatible clients such as Claude and Cursor.

Ui.Vision

Record and replay browser macros

Browser tasks are recorded and replayed with Selenium IDE-style commands: filling forms from CSV data, extracting tables and automating file downloads and uploads. Selenium IDE scripts import and export.

Computer vision and OCR

Macros can locate buttons by image and read text on screen, including inside canvas elements, images and videos, which helps where a page selector cannot be built.

Desktop automation with XModules

The free Desktop App, also called XModules, adds native mouse clicks, keyboard input and drag and drop for desktop applications, plus reading and writing macros and CSV files on disk.

AI macro assistant

You describe a task and the assistant inspects the page, writes a macro, runs it and helps fix failures. It works without signup or an API key, or with your own AI provider or a local model.

MCP bridge

An MCP bridge that runs on your computer lets an MCP-compatible AI assistant inspect the live page, create and edit macros, run them and read the logs.

Command line API

Macros can be started and monitored from the command line, so they fit into tools such as Jenkins and scripts in Python, C# or PowerShell. The XRun command calls other programs.

Skyvern

• Open source (self-hosted)$0

  The AGPL-3.0 project from GitHub, installed with pip or Docker Compose and run on your own machine or server with your own LLM provider.

• Free$0

  Skyvern Cloud with 5,000 credits (about 170 actions), 1 concurrent run, datacenter proxies, webhooks and community support.

• Hobby$29

  per month

  30,000 credits a month (about 1,200 actions), 10 concurrent runs, a basic CAPTCHA solver, stored credentials and email support.

• Pro$149

  per month

  150,000 credits a month (about 6,200 actions), 25 concurrent runs, team workspaces, 2FA and TOTP support, residential proxies and priority support.

• EnterpriseCustom

  Custom credits, 100 concurrent runs, HIPAA compliance, a SOC 2 report, custom code blocks, human-in-the-loop steps and a dedicated Slack channel.

Ui.Vision

• Free Edition$0

  Open source and free for personal and commercial use: unlimited web and desktop automation, forum support and the AI assistant within basic limits.

• AI PRO€11.90

  per month, introductory price

  10 times the free plan's AI processing limits. Shown as 50% off a regular €23.80; subscribers keep the introductory price.

• PRO EditionUS$299

  per year

  A yearly subscription that adds email support to the forum support of the Free Edition.

• Enterprise EditionUS$999

  per year, for up to 5 users

  Adds update management with separate update channels and priority support by email, phone and screen sharing. Bundles for 10 to 50 users cost more.

• Enterprise PlayerUS$1,499

  yearly license, unlimited users and machines

  Runs macros and test suites on any number of machines but cannot save new macros. Includes 10,000 OCR conversions a day.

Skyvern

No public discounts listed.

Ui.Vision

• Education

  Discounts for students and universities

  The pricing page says discounts on the paid editions are available for students and universities on request.

• Other savings

  Free Enterprise licenses for open-source projects

  Open-source projects can ask a9t9 software for a free Enterprise license, according to the pricing page.