COMPUTER USE / HUB

The evidence ledger.

Inspect claim-level sources, verification dates and source types.

181claim records

77distinct primary URLs

19qualified profiles

OpenAI Computer Use

official docs

Current docs describe browser and desktop control, code execution, and structured computer actions; GPT-6 Astra code execution is recommended.

openai-computer-use-docs · Sep 27, 2026 · verified

developers.openai.com ↗

official pricing

GPT-6 Astra Standard short-context rates are $10 input and $50 output per million tokens; other processing modes and long context differ.

openai-computer-use-pricing · Sep 27, 2026 · verified

developers.openai.com ↗

official repo

OpenAI publishes a sample application for integrating its computer-use API with execution environments.

openai-computer-use-sample · Sep 27, 2026 · verified

official docs

Control surfaces: Browser and desktop interfaces

openai-computer-use-fact-1 · Sep 27, 2026 · verified

developers.openai.com ↗

official docs

Perception: Screenshots and tool results guide the next action

openai-computer-use-fact-2 · Sep 27, 2026 · verified

developers.openai.com ↗

official docs

Action interface: Structured mouse and keyboard actions

openai-computer-use-fact-3 · Sep 27, 2026 · verified

developers.openai.com ↗

official docs

Code path: Code execution can use Playwright or PyAutoGUI

openai-computer-use-fact-4 · Sep 27, 2026 · verified

developers.openai.com ↗

official docs

Execution ownership: The developer supplies and controls action execution

openai-computer-use-fact-5 · Sep 27, 2026 · verified

developers.openai.com ↗

official repo

Integration reference: Official sample application is publicly available

openai-computer-use-fact-6 · Sep 27, 2026 · verified

Claude Computer Use

official docs

computer_toolset_20260801 provides screenshot and input tools; actions execute in the client environment. Documentation warns about latency and security.

claude-computer-use-docs · Sep 27, 2026 · verified

platform.claude.com ↗

official pricing

Claude Sonnet 5 base pricing is $2 per million input tokens and $10 per million output tokens; computer tool definitions add input overhead.

claude-computer-use-pricing · Sep 27, 2026 · verified

platform.claude.com ↗

official repo

Anthropic maintains computer-use quickstart code in its public quickstarts repository.

claude-computer-use-sample · Sep 27, 2026 · verified

official docs

Desktop input: Mouse and keyboard actions in a desktop environment

claude-computer-use-fact-1 · Sep 27, 2026 · verified

platform.claude.com ↗

official docs

Screen perception: Screenshot and zoom tool members

claude-computer-use-fact-2 · Sep 27, 2026 · verified

platform.claude.com ↗

official docs

Batch actions: Ordered short action sequences can share a response

claude-computer-use-fact-3 · Sep 27, 2026 · verified

platform.claude.com ↗

official docs

Execution loop: Client-side tool handlers execute requested actions

claude-computer-use-fact-4 · Sep 27, 2026 · verified

platform.claude.com ↗

official docs

Environment reference: The reference setup uses a sandboxed Linux desktop

claude-computer-use-fact-5 · Sep 27, 2026 · verified

platform.claude.com ↗

official docs

Toolset version: computer_toolset_20260801; compatibility is model-specific

claude-computer-use-fact-6 · Sep 27, 2026 · verified

platform.claude.com ↗

Gemini Computer Use

official docs

Current guide covers browser, desktop and mobile environments, client execution, safety decisions, and Gemini 3.8 Flash as the recommended model.

gemini-computer-use-docs · Sep 27, 2026 · verified

ai.google.dev ↗

official pricing

Gemini pricing is model-, processing-, and context-dependent. This snapshot deliberately omits numeric Gemini rates until the exact variant is normalized.

gemini-computer-use-pricing · Sep 27, 2026 · verified

ai.google.dev ↗

official repo

Google publishes a reference computer-use agent implementation; examples require client-side browser execution.

gemini-computer-use-sample · Sep 27, 2026 · verified

official docs

Environment coverage: Browser, mobile and desktop control agents

gemini-computer-use-fact-1 · Sep 27, 2026 · verified

ai.google.dev ↗

official docs

Visual input: Screenshots provide the model's current UI state

gemini-computer-use-fact-2 · Sep 27, 2026 · verified

ai.google.dev ↗

official docs

UI action output: Function calls describe clicks, scrolling and keystrokes

gemini-computer-use-fact-3 · Sep 27, 2026 · verified

ai.google.dev ↗

official docs

Safety response: Actions may require confirmation or be blocked

gemini-computer-use-fact-4 · Sep 27, 2026 · verified

ai.google.dev ↗

official docs

Execution ownership: Client code executes approved actions and returns screenshots

gemini-computer-use-fact-5 · Sep 27, 2026 · verified

ai.google.dev ↗

official repo

Reference implementation: Public example includes browser execution code

gemini-computer-use-fact-6 · Sep 27, 2026 · verified

Browser Use

official docs

API V4 supports hosted runs and browser sessions; Python and TypeScript examples show CDP connections and explicit browser stop.

browser-use-docs · Sep 27, 2026 · verified

docs.browser-use.com ↗

official repo

The maintained open-source library implements model-driven browser agents and supports local browser execution.

browser-use-repo · Sep 27, 2026 · verified

official pricing

Browsers cost $0.02/hour plus traffic; residential proxies $5/GB; managed agents add 20% to model cost.

browser-use-pricing · Sep 27, 2026 · verified

browser-use.com ↗

official docs

Browser Use supports multiple model providers and custom OpenAI-compatible endpoints; coordinate clicking is model-dependent.

browser-use-models · Sep 27, 2026 · verified

docs.browser-use.com ↗

official docs

Agent parameters explicitly document screenshot vision, browser state/DOM, coordinate and element clicking, typing and keyboard shortcuts.

browser-use-parameters · Sep 27, 2026 · verified

docs.browser-use.com ↗

official docs

Managed tasks: Hosted runs accept a task and expose completion results

browser-use-fact-1 · Sep 27, 2026 · verified

docs.browser-use.com ↗

official docs

Cloud browser: Launch a browser and connect over CDP

browser-use-fact-2 · Sep 27, 2026 · verified

docs.browser-use.com ↗

official docs

SDK languages: Python and TypeScript SDK quickstarts

browser-use-fact-3 · Sep 27, 2026 · verified

docs.browser-use.com ↗

official repo

Local agent: Open-source library supports a local agent path

browser-use-fact-4 · Sep 27, 2026 · verified

official docs

Browser observation: Configurable screenshots and DOM state guide agent actions

browser-use-fact-5 · Sep 27, 2026 · verified

docs.browser-use.com ↗

official docs

Session lifecycle: Stop the managed browser explicitly after a run

browser-use-fact-6 · Sep 27, 2026 · verified

docs.browser-use.com ↗

Skyvern

official docs

Skyvern provides a developer quickstart for browser task execution and its SDK.

skyvern-docs · Sep 27, 2026 · verified

www.skyvern.com ↗

official repo

Repository documents vision-based actions, Playwright-compatible SDK, no-code workflows, schema extraction, file downloads and MCP; AGPL-3.0.

skyvern-repo · Sep 27, 2026 · verified

official pricing

Hobby costs $29/month with 30,000 credits; includes webhooks, stored credentials and country targeting.

skyvern-pricing · Sep 27, 2026 · verified

www.skyvern.com ↗

official repo

Visual action selection: Vision models interpret websites to select actions

skyvern-fact-1 · Sep 27, 2026 · verified

official repo

Developer interface: Playwright-compatible SDK adds AI actions

skyvern-fact-2 · Sep 27, 2026 · verified

official repo

Workflow composition: Chain tasks, validation and loops in workflows

skyvern-fact-3 · Sep 27, 2026 · verified

official repo

Structured extraction: Supply an output data schema

skyvern-fact-4 · Sep 27, 2026 · verified

official repo

File handling: Browser downloads can be uploaded to configured storage

skyvern-fact-5 · Sep 27, 2026 · verified

official repo

Agent connectivity: MCP integration is documented

skyvern-fact-6 · Sep 27, 2026 · verified

Stagehand

official docs

Stagehand v4 combines act/extract/observe with Playwright-style page APIs and TypeScript, Python and Go support.

stagehand-docs · Sep 27, 2026 · verified

docs.stagehand.dev ↗

official repo

Stagehand is a public SDK repository with local and Browserbase integration paths.

stagehand-repo · Sep 27, 2026 · verified

official product

Stagehand is presented as a browser agent SDK with agentic and programmatic control.

stagehand-product · Sep 27, 2026 · verified

www.stagehand.dev ↗

official docs

Stagehand documents provider and model configuration for developer-selected models.

stagehand-models · Sep 27, 2026 · verified

docs.stagehand.dev ↗

official docs

Natural-language actions: act translates a requested browser step into action

stagehand-fact-1 · Sep 27, 2026 · verified

docs.stagehand.dev ↗

official docs

Schema extraction: extract returns structured data

stagehand-fact-2 · Sep 27, 2026 · verified

docs.stagehand.dev ↗

official docs

Action discovery: observe identifies available actions

stagehand-fact-3 · Sep 27, 2026 · verified

docs.stagehand.dev ↗

official docs

Deterministic control: Page APIs include goto, click, type and locators

stagehand-fact-4 · Sep 27, 2026 · verified

docs.stagehand.dev ↗

official docs

SDK languages: TypeScript, Python and Go

stagehand-fact-5 · Sep 27, 2026 · verified

docs.stagehand.dev ↗

official repo

Deployment choice: Local and managed browser integration paths

stagehand-fact-6 · Sep 27, 2026 · verified

Playwright

official docs

Playwright Test bundles assertions, isolation, parallelization and tooling; supports Windows, Linux and macOS with headed/headless execution.

playwright-docs · Sep 27, 2026 · verified

playwright.dev ↗

official docs

Chromium, Firefox and WebKit are supported; mobile devices can be emulated and browser binaries must match the release.

playwright-browsers · Sep 27, 2026 · verified

playwright.dev ↗

official repo

Microsoft maintains Playwright as a public Apache-2.0 browser automation project.

playwright-repo · Sep 27, 2026 · verified

official docs

The screenshot API captures full pages or individual elements and can return a buffer.

playwright-screenshots · Sep 27, 2026 · verified

playwright.dev ↗

official docs

Browser engines: Chromium, Firefox and WebKit

playwright-fact-1 · Sep 27, 2026 · verified

playwright.dev ↗

official docs

Operating systems: Windows, Linux and macOS

playwright-fact-2 · Sep 27, 2026 · verified

playwright.dev ↗

official docs

Execution modes: Headless or headed, locally or in CI

playwright-fact-3 · Sep 27, 2026 · verified

playwright.dev ↗

official docs

Test facilities: Assertions, isolated contexts and parallel execution

playwright-fact-4 · Sep 27, 2026 · verified

playwright.dev ↗

official docs

Device emulation: Mobile browser emulation is supported

playwright-fact-5 · Sep 27, 2026 · verified

playwright.dev ↗

official repo

Distribution: Apache-2.0 public source repository

playwright-fact-6 · Sep 27, 2026 · verified

Selenium

official docs

Selenium documents WebDriver browser control, IDE test recording and Grid remote execution.

selenium-docs · Sep 27, 2026 · verified

www.selenium.dev ↗

official docs

Grid supports remote parallel browser sessions across machines, versions and operating systems.

selenium-grid · Sep 27, 2026 · verified

www.selenium.dev ↗

official repo

Public Selenium source includes language bindings and the server under Apache-2.0.

selenium-repo · Sep 27, 2026 · verified

official docs

Browser control: WebDriver drives browser interactions

selenium-fact-1 · Sep 27, 2026 · verified

www.selenium.dev ↗

official docs

Recorded tests: Selenium IDE records and plays back browser tests

selenium-fact-2 · Sep 27, 2026 · verified

www.selenium.dev ↗

official docs

Remote execution: Grid routes commands to remote browser instances

selenium-fact-3 · Sep 27, 2026 · verified

www.selenium.dev ↗

official docs

Parallel coverage: Run across browser versions and operating systems

selenium-fact-4 · Sep 27, 2026 · verified

www.selenium.dev ↗

official repo

Language bindings: Client libraries support multiple programming languages

selenium-fact-5 · Sep 27, 2026 · verified

official repo

Distribution: Apache-2.0 public source repository

selenium-fact-6 · Sep 27, 2026 · verified

Playwright MCP

official repo

MCP server uses accessibility-tree observations, configurable browsers, profiles and optional vision tools; maintainer states it is not a security boundary.

playwright-mcp-repo · Sep 27, 2026 · verified

official docs

The Playwright project publishes MCP setup guidance for installing the server into a compatible client.

playwright-mcp-guide · Sep 27, 2026 · verified

official repo

The Playwright MCP source is licensed under Apache License 2.0.

playwright-mcp-license · Sep 27, 2026 · verified

official repo

Protocol: MCP tools connect a compatible agent client to the browser

playwright-mcp-fact-1 · Sep 27, 2026 · verified

official repo

Primary observation: Structured accessibility snapshots

playwright-mcp-fact-2 · Sep 27, 2026 · verified

official repo

Visual mode: Optional vision capability can be enabled

playwright-mcp-fact-3 · Sep 27, 2026 · verified

official repo

Profiles: Persistent, isolated or extension-connected browser sessions

playwright-mcp-fact-4 · Sep 27, 2026 · verified

official repo

Browser selection: Chrome, Firefox, WebKit and Edge configuration

playwright-mcp-fact-5 · Sep 27, 2026 · verified

official repo

License: Apache-2.0

playwright-mcp-fact-6 · Sep 27, 2026 · verified

Browserbase

official docs

Browserbase provides managed browsers for automation with existing browser frameworks and session inspection.

browserbase-docs · Sep 27, 2026 · verified

docs.browserbase.com ↗

official product

Browserbase offers cloud browser execution for computer-use model integrations.

browserbase-product · Sep 27, 2026 · verified

www.browserbase.com ↗

official pricing

Developer plan: $20/month, 100 included browser hours then $0.12/hour; 1 GB proxy allowance then $12/GB; 25 concurrent browsers.

browserbase-pricing · Sep 27, 2026 · verified

www.browserbase.com ↗

official docs

Browserbase documents screenshot capture from connected Playwright and Puppeteer sessions.

browserbase-screenshots · Sep 27, 2026 · verified

docs.browserbase.com ↗

official docs

Managed execution: Launch browser sessions through an API

browserbase-fact-1 · Sep 27, 2026 · verified

docs.browserbase.com ↗

official docs

Framework integration: Connect existing browser automation tools

browserbase-fact-2 · Sep 27, 2026 · verified

docs.browserbase.com ↗

official product

Computer-use hosting: Run model-driven interactions in cloud browsers

browserbase-fact-3 · Sep 27, 2026 · verified

www.browserbase.com ↗

official docs

Session inspection: Observe and debug browser sessions

browserbase-fact-4 · Sep 27, 2026 · verified

docs.browserbase.com ↗

official pricing

Proxy capacity: Plan-based proxy bandwidth allowance and overages

browserbase-fact-5 · Sep 27, 2026 · verified

www.browserbase.com ↗

official pricing

Concurrency: Developer plan documents 25 concurrent browsers

browserbase-fact-6 · Sep 27, 2026 · verified

www.browserbase.com ↗

Steel

official product

Steel documents session APIs, context reuse, Playwright/Puppeteer/Selenium connectivity, live sessions and replay.

steel-product · Sep 27, 2026 · verified

official repo

Steel Browser is a public browser sandbox/API implementation for local or self-hosted deployment.

steel-repo · Sep 27, 2026 · verified

official pricing

Launch is $0 plus usage; browser time $0.10/hour and proxy bandwidth $10/GB; 15-minute sessions and one-time $30 credit. Pricing page last edit June 30, 2026.

steel-pricing · Sep 27, 2026 · verified

docs.steel.dev ↗

official docs

Browser Tools provides a screenshot endpoint returning a hosted PNG, with optional full-page capture; this is separate from long-lived sessions.

steel-browser-tools · Sep 27, 2026 · verified

docs.steel.dev ↗

official product

Session API: Create on-demand cloud browser sessions

steel-fact-1 · Sep 27, 2026 · verified

official product

Context reuse: Save and inject cookies and local storage

steel-fact-2 · Sep 27, 2026 · verified

official product

Framework support: Puppeteer, Playwright and Selenium integrations

steel-fact-3 · Sep 27, 2026 · verified

official product

Observability: View live or recorded sessions

steel-fact-4 · Sep 27, 2026 · verified

official repo

Runtime source: Public browser runtime supports self-hosting

steel-fact-5 · Sep 27, 2026 · verified

official pricing

Proxy meter: Proxy usage is billed separately from browser time

steel-fact-6 · Sep 27, 2026 · verified

docs.steel.dev ↗

Hyperbrowser

official docs

Documentation lists cloud browser sessions, web APIs, AI agents, Playwright/Puppeteer integration, proxies, and Python/Node SDKs.

hyperbrowser-docs · Sep 27, 2026 · verified

www.hyperbrowser.ai ↗

official repo

Node SDK example creates a session, connects Playwright over CDP, navigates pages, and stops the session in finally.

hyperbrowser-sdk · Sep 27, 2026 · verified

official repo

Hyperbrowser maintains an MCP server for access from compatible AI clients.

hyperbrowser-mcp · Sep 27, 2026 · verified

official docs

Managed sessions: Create and manage cloud browser sessions

hyperbrowser-fact-1 · Sep 27, 2026 · verified

www.hyperbrowser.ai ↗

official docs

Existing scripts: Playwright and Puppeteer integration

hyperbrowser-fact-2 · Sep 27, 2026 · verified

www.hyperbrowser.ai ↗

official docs

SDKs: Python and Node SDKs are documented

hyperbrowser-fact-3 · Sep 27, 2026 · verified

www.hyperbrowser.ai ↗

official repo

Direct connection: Node SDK exposes a CDP WebSocket endpoint

hyperbrowser-fact-4 · Sep 27, 2026 · verified

official repo

Agent tools: Official MCP server integration

hyperbrowser-fact-5 · Sep 27, 2026 · verified

official repo

Session cleanup: SDK examples explicitly stop sessions after use

hyperbrowser-fact-6 · Sep 27, 2026 · verified

Anchor Browser

official docs

Anchor provides cloud navigation, authentication, form filling, extraction, and Python/Node task examples.

anchor-browser-docs · Sep 27, 2026 · verified

docs.anchorbrowser.io ↗

official product

Anchor presents infrastructure for computer-use agents and authenticated browser automation.

anchor-browser-product · Sep 27, 2026 · verified

anchorbrowser.io ↗

official docs

Agent Access provides programmatic API-key onboarding and documents a 60-minute session limit.

anchor-browser-access · Sep 27, 2026 · verified

docs.anchorbrowser.io ↗

official docs

Browser hosting: A real cloud browser for agent actions

anchor-browser-fact-1 · Sep 27, 2026 · verified

docs.anchorbrowser.io ↗

official docs

Web actions: Navigation, form filling and extraction

anchor-browser-fact-2 · Sep 27, 2026 · verified

docs.anchorbrowser.io ↗

official docs

Task API: Natural-language task invocation

anchor-browser-fact-3 · Sep 27, 2026 · verified

docs.anchorbrowser.io ↗

official docs

SDK examples: Python and Node integration

anchor-browser-fact-4 · Sep 27, 2026 · verified

docs.anchorbrowser.io ↗

official product

Authentication use: Authenticated website workflows are a product focus

anchor-browser-fact-5 · Sep 27, 2026 · verified

anchorbrowser.io ↗

official docs

Programmatic onboarding: Agent Access can issue scoped onboarding credentials

anchor-browser-fact-6 · Sep 27, 2026 · verified

docs.anchorbrowser.io ↗

Browserless

official docs

Browserless documents browser automation APIs, BrowserQL and connections for browser frameworks.

browserless-docs · Sep 27, 2026 · verified

docs.browserless.io ↗

official repo

Deployable Docker browser infrastructure; repository states free for non-commercial uses.

browserless-repo · Sep 27, 2026 · verified

official pricing

Prototyping is $25/month billed annually, with 20,000 units and $0.0020/unit overages; plan includes reconnects and multiple browser engines.

browserless-pricing · Sep 27, 2026 · verified

www.browserless.io ↗

official docs

Hosted browsers: Cloud browser execution through APIs

browserless-fact-1 · Sep 27, 2026 · verified

docs.browserless.io ↗

official repo

Deployable runtime: Browser infrastructure can run in Docker

browserless-fact-2 · Sep 27, 2026 · verified

official pricing

BrowserQL: Browser-oriented automation language and editor

browserless-fact-3 · Sep 27, 2026 · verified

www.browserless.io ↗

official pricing

Session continuity: Reconnect and persisted-session features

browserless-fact-4 · Sep 27, 2026 · verified

www.browserless.io ↗

official pricing

Browser engines: Chrome, WebKit and Firefox listed in plan features

browserless-fact-5 · Sep 27, 2026 · verified

www.browserless.io ↗

official pricing

Network options: Residential, datacenter and bring-your-own proxy options vary by plan

browserless-fact-6 · Sep 27, 2026 · verified

www.browserless.io ↗

Kernel

official product

Kernel describes isolated Chromium, CDP/WebDriver BiDi, profiles, managed auth, SDKs, MCP and live/replayed observation.

kernel-product · Sep 27, 2026 · verified

www.kernel.sh ↗

official docs

Documentation details computer controls, Playwright execution, screenshots, profiles and cloud browser sessions.

kernel-docs · Sep 27, 2026 · verified

www.kernel.sh ↗

official pricing

Developer plan is free plus usage. Headless browser costs $0.0000166667/sec, headful $0.0001333336/sec; proxies are not separately charged.

kernel-pricing · Sep 27, 2026 · verified

www.kernel.sh ↗

official docs

Isolation: Sandboxed cloud Chromium sessions

kernel-fact-1 · Sep 27, 2026 · verified

www.kernel.sh ↗

official docs

Control options: Computer controls, Playwright execution, CDP and WebDriver BiDi

kernel-fact-2 · Sep 27, 2026 · verified

www.kernel.sh ↗

official docs

Observation: Live view, screenshots and MP4 replay

kernel-fact-3 · Sep 27, 2026 · verified

www.kernel.sh ↗

official product

Persistent state: Profiles preserve browser state between sessions

kernel-fact-4 · Sep 27, 2026 · verified

www.kernel.sh ↗

official docs

Authentication: Managed Auth and vault-based login options

kernel-fact-5 · Sep 27, 2026 · verified

www.kernel.sh ↗

official product

Agent integration: Official SDKs, CLI and hosted MCP

kernel-fact-6 · Sep 27, 2026 · verified

www.kernel.sh ↗

Airtop

official docs

Cloud sessions, natural-language AI APIs, paginated extraction, visual/text/DOM observation and Playwright/Puppeteer/Selenium integration are documented.

airtop-docs · Sep 27, 2026 · verified

docs.airtop.ai ↗

official product

Airtop offers agent compilation, schedules, API integrations and browser operations behind login.

airtop-product · Sep 27, 2026 · verified

www.airtop.ai ↗

official pricing

Starter is listed at $29/month with three simultaneous sessions; Free lists 1,000 credits and one deployed agent.

airtop-pricing · Sep 27, 2026 · verified

www.airtop.ai ↗

official docs

Cloud sessions: Create a browser session from the SDK

airtop-fact-1 · Sep 27, 2026 · verified

docs.airtop.ai ↗

official docs

AI control: Natural-language browser API

airtop-fact-2 · Sep 27, 2026 · verified

docs.airtop.ai ↗

official docs

Extraction: Paginated structured extraction

airtop-fact-3 · Sep 27, 2026 · verified

docs.airtop.ai ↗

official docs

Observation: Combines visual, text and DOM information

airtop-fact-4 · Sep 27, 2026 · verified

docs.airtop.ai ↗

official docs

Script integration: Playwright, Puppeteer and Selenium support

airtop-fact-5 · Sep 27, 2026 · verified

docs.airtop.ai ↗

official product

Scheduled workflows: Agent Builder can compile and schedule repeatable automations

airtop-fact-6 · Sep 27, 2026 · verified

www.airtop.ai ↗

Agent S by Simular

official repo

Agent S is an open-source framework for screenshot-based clicking, typing and scrolling on macOS, Windows and Linux with multiple model providers.

simular-repo · Sep 27, 2026 · verified

official product

Simular distinguishes its Agent S open-source research from hosted Sai and business products.

simular-product · Sep 27, 2026 · verified

www.simular.ai ↗

official product

Simular's company page identifies its computer-agent research focus and the team behind the framework.

simular-about · Sep 27, 2026 · verified

www.simular.ai ↗

official repo

Task input: Natural-language instructions drive computer tasks

simular-fact-1 · Sep 27, 2026 · verified

official repo

Perception: Screen observations guide the agent

simular-fact-2 · Sep 27, 2026 · verified

official repo

Actions: Click, type and scroll in desktop and web applications

simular-fact-3 · Sep 27, 2026 · verified

official repo

Platforms: macOS, Windows and Linux

simular-fact-4 · Sep 27, 2026 · verified

official repo

Model choice: OpenAI, Anthropic and open-weight provider paths

simular-fact-5 · Sep 27, 2026 · verified

official product

Project scope: Open-source framework is separate from hosted Sai

simular-fact-6 · Sep 27, 2026 · verified

www.simular.ai ↗

Browse AI

official product

Browse AI provides no-code data extraction, prebuilt robots, integrations, API access and website monitoring.

browse-ai-product · Sep 27, 2026 · verified

www.browse.ai ↗

official product

Train an agent on a page, select data such as text or screenshots, and add a scheduled monitor.

browse-ai-monitor · Sep 27, 2026 · verified

www.browse.ai ↗

official pricing

Personal annual view is $19/month billed annually with five websites, three users and unlimited robots; credit limits apply.

browse-ai-pricing · Sep 27, 2026 · verified

www.browse.ai ↗

official product

No-code extraction: Train a robot to collect website data

browse-ai-fact-1 · Sep 27, 2026 · verified

www.browse.ai ↗

official product

Change monitoring: Add monitors to trained robots

browse-ai-fact-2 · Sep 27, 2026 · verified

www.browse.ai ↗

official product

Target data: Monitor text, rank, image or screenshot changes

browse-ai-fact-3 · Sep 27, 2026 · verified

www.browse.ai ↗

official product

Integration paths: Product provides integrations and API access

browse-ai-fact-4 · Sep 27, 2026 · verified

www.browse.ai ↗

official product

Repeat runs: Schedule monitoring instead of manually revisiting pages

browse-ai-fact-5 · Sep 27, 2026 · verified

www.browse.ai ↗

official pricing

Team allowance: Personal plan lists three users and five websites

browse-ai-fact-6 · Sep 27, 2026 · verified

www.browse.ai ↗

Axiom

official product

Axiom supports visual browser steps, AI-assisted authoring, code APIs and cloud or desktop execution.

axiom-product · Sep 27, 2026 · verified

official docs

Documentation covers no-code automation, Chrome/Step/Orchestrate APIs, scheduling and integrations.

axiom-docs · Sep 27, 2026 · verified

official pricing

Starter costs $15/month for five runtime hours; one cloud bot and a one-hour cloud single-run limit. API and MCP access differ by plan.

axiom-pricing · Sep 27, 2026 · verified

official product

Visual authoring: Compose browser automation steps in a Chrome extension

axiom-fact-1 · Sep 27, 2026 · verified

official product

AI assistance: Build automations from a description

axiom-fact-2 · Sep 27, 2026 · verified

official docs

Code interfaces: Chrome, Step and Orchestrate APIs

axiom-fact-3 · Sep 27, 2026 · verified

official product

Execution choice: Run on the vendor cloud or your own machine

axiom-fact-4 · Sep 27, 2026 · verified

official pricing

Agent integration: MCP server is offered on qualifying plans

axiom-fact-5 · Sep 27, 2026 · verified

official pricing

Login handling: Saved logins and TOTP support are listed in plans

axiom-fact-6 · Sep 27, 2026 · verified

Benchmark methodologies

academic

Desktop task evaluation in real computer environments. Older reported scores require their original step budget, observation and harness settings.

Verified Sep 27, 2026 · OSWorld 1.0 results are not comparable to OSWorld V2 scores; this atlas has not normalized a numeric 1.0 result.

Original methodology ↗

academic

Versioned long-horizon desktop evaluation. The repository recommends the 2.1 release and matching code, task assets, websites and environment images.

Verified Sep 27, 2026 · No score is imported until its exact release and harness can be matched. Do not substitute main/latest for a versioned benchmark release.

Original methodology ↗

official product

Browser Use reports a matched evaluation of browser providers across protected websites using Halluminate's BrowserBench workload.

Verified Sep 27, 2026 · Vendor-run access benchmark, not end-to-end agent task success. Upstream commit, provider settings and raw run date are not fully disclosed; compare only within this reported snapshot.

Original methodology ↗

official product

A vendor-run set of 106 live-web tasks. A task earns success only on an exact final answer or end state; cost is total recorded spend divided by solved tasks.

Verified Sep 27, 2026 · Private workload; model labels and harness details are reported by the vendor. This is not an independent ranking and does not transfer to current model versions.

Original methodology ↗

Editorial primary-source registry

Reviewed 2026-09-27. These original documentation and methodology pages support the editorial guides and workflow architecture recommendations.

RESEARCH NOTES

Sources & verification

Reviewed Sep 27, 2026. Architecture recommendations are editorial analysis; linked vendor documentation supports the underlying capability and safety facts.

How we verify evidence →