Comparación de agentes de IA para trabajo general
Agentes que operan dentro de aplicaciones e integraciones, y entre ellas, para ejecutar una amplia variedad de tareas de conocimiento. Pueden completar tareas de varios pasos de forma autónoma en un espacio de trabajo delegado durante periodos prolongados (por ejemplo, leer y escribir archivos, acceder a funciones del ordenador y del navegador, y coordinarse con subagentes).
Para comparar modelos de lenguaje, consulta nuestros benchmarks de modelos.
| Product | Released | Platform | Open Source | Bring Your Own Model | Local Files | Browser Automation | Price | Description | |
|---|---|---|---|---|---|---|---|---|---|
| Claude Cowork | Jan 2026 | macOSWindows | SubscriptionEnterprise $20–200/mo | Desktop-first agent that executes tasks in a sandboxed Linux VM with access to local files. The agent can edit and create local files, and has pre-set skills for handling common file types and general work tasks (e.g. editing Excel spreadsheets). Connects to external tools via computer use, MCP connectors, or browser extension. | |||||
| ChatGPT Agent | Jul 2025 | WebMobileDesktop | SubscriptionEnterprise $20–200/mo | Agentic system within the ChatGPT app, powered by GPT-5.4 Thinking/Pro — OpenAI's frontier model for agentic workflows. It runs on a cloud VM with no direct file access. GPT-5.4 brings native computer-use capabilities and desktop-level automation, designed for longer-horizon tasks with up to ~30+ minutes of autonomous execution. Successor to Operator and combines Codex-level coding and Deep Research into one experience. | |||||
| OpenClaw | Nov 2025 | macOSLinuxWindows | Free Free + API costs | Open-source, self-hosted agent platform that turns existing chat apps (e.g. WhatsApp, Slack, Teams) into autonomous AI assistants. Also has macOS menu bar companion app and WebChat interface. Tasks run in the background with results in chat. Supports community Skills marketplace, NVIDIA NemoClaw stack for secure scaling, and security models. | |||||
| Hermes Agent | Feb 2026 | macOSLinuxWindows | Free Free + API costs | Open-source, self-hosted agent platform with persistent cross-session memory and a self-improving skills system. Creates procedural skills from experience and reuses them across sessions. Tasks run in the background with results in chat. Includes 40+ built-in tools (e.g. image gen, TTS, vision, Telegram, Discord, Slack, WhatsApp). | |||||
| Microsoft Copilot | Nov 2023 | WindowsmacOSWebiOSAndroid | SubscriptionEnterpriseUsage-based $9.99–30/mo | Agents run inside Microsoft 365 ecosystem (Teams, Word, Excel, PowerPoint, and Outlook), executing multi-step tasks like document creation, data analysis, and scheduling. Connects to external tools via MCP, Power Platform connectors, and Microsoft Graph. Copilot Studio lets organizations build and deploy custom autonomous agents. | |||||
| Google Workspace Studio | Dec 2025 | Web | SubscriptionEnterprise ~$8–27/mo | No-code agent builder inside Google Workspace for creating and sharing AI agents. Agents reason, adapt to new information, and handle tasks like email summarization, meeting action items, and document management, with deep Google ecosystem integration (Gmail, Drive, Sheets, Maps, Docs, etc.) and external app support via Marketplace connectors. | |||||
| Gemini Enterprise | Oct 2025 | Web | FreeSubscriptionEnterprise $21–30/mo | Agentic platform for enterprise knowledge workers. Connects to company data via prebuilt connectors (M365, Salesforce, Jira, Confluence, Box). Offers permission-aware search, prebuilt agents like Deep Research, no-code workbench, and Code Assist. | |||||
| Manus | Mar 2025 | WebWindowsmacOSiOSAndroid | FreeSubscriptionUsage-based $20–200/mo | General purpose agent that can orchestrate multiple AI models per task and autonomously plan and execute (web automation, coding, data work, PDF/PPT generation) in a sandboxed environment. Produces deliverables including research reports, websites, and data analysis. | |||||
| Microsoft Copilot Cowork | Mar 2026 | Web | SubscriptionEnterpriseUsage-based $21–30/mo | Execution layer for Microsoft 365 Copilot that handles autonomous, multi-step actions across Outlook, Teams, Excel, Word, PowerPoint, and Work IQ. Fully sandboxed your M365 cloud tenant, permission-aware, and enterprise-compliant. Plans, coordinates, and executes tasks (calendar cleanup, competitive research, launch plans, etc.) in the background with checkpoints, clarification requests, and user approvals. |
Resumen del panorama
La tabla compara agentes de trabajo general según disponibilidad de plataforma, estado open-source, soporte para traer tu propio modelo, acceso a archivos locales, automatización del navegador y precios. Microsoft Copilot y Manus ofrecen el soporte de plataformas más amplio (desktop + web + móvil), mientras que ChatGPT Agent cubre web, móvil y desktop. Claude Cowork y OpenClaw son solo de desktop. Google Workspace Studio, Gemini Enterprise y Microsoft Copilot Cowork son solo web. La mayoría de los agentes admiten acceso a archivos locales; ChatGPT Agent es la excepción destacada, ya que se ejecuta por completo en una VM en la nube. La automatización del navegador está disponible en Claude Cowork, ChatGPT Agent, OpenClaw, Manus y parcialmente en Microsoft Copilot. Los precios van desde opciones open-source gratuitas (OpenClaw) hasta suscripciones de $8–30/mes (Google Workspace Studio, Microsoft Copilot, Gemini Enterprise) y niveles de hasta $200/mes (Claude Cowork, ChatGPT Agent, Manus).
Preguntas frecuentes
General work AI agents are AI systems that autonomously execute multi-step tasks over extended periods. They read and write files on your desktop or in cloud sandboxes, perform broad knowledge work tasks, and typically connect to external services like email, calendars, Slack, and CRM. Unlike coding agents, they target everyday office and administrative tasks for non-developer knowledge workers.
Most agents in the table support local file access: Claude Cowork, OpenClaw, Microsoft Copilot, Google Workspace Studio, Manus, and Microsoft Copilot Cowork. ChatGPT Agent is the notable exception — it runs entirely on a cloud VM with no local file access. Gemini Enterprise offers partial access via connectors to SharePoint, Confluence, and other enterprise systems.
Claude Cowork, ChatGPT Agent, OpenClaw, and Manus all support browser automation. Microsoft Copilot offers partial browser automation via Copilot Studio. The remaining agents (Google Workspace Studio, Gemini Enterprise, Microsoft Copilot Cowork) do not include browser automation.
OpenClaw is fully open-source and free (you pay only for LLM API calls). Microsoft Copilot has a partial BYOM option via Copilot Studio. The remaining agents are closed-source and require subscriptions ranging from ~$8/mo (Google Workspace Studio) to $200/mo (Claude Cowork, ChatGPT Agent).
We compare agents across platform availability, open-source status, bring-your-own-model support, local file access, browser automation, and pricing. Our comparison table is updated regularly. View LLM benchmarks