Compare · Cloud agent · Checked November 9, 2026

Hermes Agentvs Manus

Hermes Agent is open source and runs on your machine or Hermes Cloud, with the model you pick. Manus is closed source, runs tasks in its cloud sandbox by default and bills credits per task.

Key differencesHermes AgentManus
Source code
Open source (MIT)
Closed source
Where it runs
Your machines or Hermes Cloud
Manus's cloud, plus desktop app
Models
Hundreds, plus local models
Manus picks, or your own key
Chat apps
More than 20 platforms
Telegram, LINE, Slack, email

Why Hermes Agent

  1. [ 01 ]

    Open source under the MIT license

    The code is on GitHub, and the MIT license lets you change it and use it commercially. Manus is closed source.

  2. [ 02 ]

    Runs on your hardware or Hermes Cloud

    Run it on your computer, a server or Hermes Cloud; paid Nous Portal plans include one Cloud agent. Manus runs tasks in a cloud sandbox by default.

  3. [ 03 ]

    The model you pick is the model that answers

    Pick frontier models on launch day on Nous Portal, any OpenAI-compatible API or a local model. Manus picks the model for each task by default.

  4. [ 04 ]

    Answers on more than 20 messaging platforms

    One gateway connects WhatsApp, Signal, Telegram, Discord, Slack, Microsoft Teams, LINE and more. Manus Agents work in Telegram, LINE and Slack.

At a glance

A check marks where Hermes Agent has the edge.
Hermes Agent and Manus compared by topic
TopicHermes AgentManus
Source codeHermes Agent:Edge: Open source under the MIT license, on GitHub.Manus:Closed source. Its terms forbid reverse engineering the service.
Where it runsHermes Agent:Edge: Your computer, a server or Hermes Cloud.Manus:A temporary cloud sandbox by default, or a folder via its desktop app.
ModelsHermes Agent:Edge: Hundreds of Nous Portal models, any OpenAI-compatible API or local models.Manus:Manus picks by default. Manus Flex takes your own key from partners.
Memory and skillsHermes Agent:Edge: Curated memory, searchable past sessions, and skills it writes itself.Manus:Projects share instructions and files. Tasks can be saved as Skills.
Chat appsHermes Agent:Edge: More than 20 messaging platforms, including WhatsApp, Signal and Teams.Manus:Telegram, LINE and Slack, plus tasks by email.
Your dataHermes Agent:Edge: Stored on the machine that runs it. No telemetry by default.Manus:Stored in the US and Singapore, including sandbox files and task logs.
PriceHermes Agent:Free software. Pay your model provider, or use a Nous Portal plan.Manus:Credits per task. Free daily credits work only in Lite mode.

Switch from Manus

  1. Install Hermes Agent

    Get Hermes Desktop for macOS (Apple Silicon), Windows and Linux, or the one-line installer on a server or WSL2.

  2. Sign in to Nous Portal

    Run hermes setup --portal, or sign in with one click in Hermes Desktop, to connect hundreds of models.

  3. Connect your chat apps

    Run hermes gateway setup to reach your agent from Telegram, Slack, WhatsApp and other chat apps.

hermes setup --portal

In detail

Open source, on your hardware or Hermes Cloud

Manus is a closed-source general AI agent from Butterfly Effect Pte. Ltd., a Singapore company. By default it runs each task in a temporary cloud sandbox, and its terms for individual users forbid reverse engineering the service. Hermes Agent is open source under the MIT license, which allows commercial use. Its code is on GitHub, and you can modify it or add tools through its plugin system.

Hermes Agent runs on your computer, a server or a VPS. Its shell commands can run locally, in a Docker container, over SSH, or in Modal, Daytona or Vercel Sandbox. If you do not want to keep a machine on, Hermes Cloud hosts your agent on a dedicated, always-on instance in the nearest available region. Every paid Nous Portal plan includes one Hermes Cloud agent; larger or extra agents bill daily from credits. Its files, memory and configuration persist whether it is running or stopped.

Manus also offers a Cloud Computer, a persistent virtual machine you can purchase for projects that need to keep running, and its desktop app, Manus Studio, can work in a folder you choose on a Mac or Windows PC.

The model you pick is the model that answers

Hermes Agent works with any OpenAI-compatible API. One Nous Portal login covers hundreds of models and hosted tools, and the provider list includes OpenRouter and dozens of others. A ChatGPT sign-in goes through the Codex provider and uses Codex models; SuperGrok and X Premium+ subscribers can also sign in. Local models run through Ollama, vLLM, llama.cpp or SGLang with no account.

You pick the provider and the model, and /model switches mid-session while conversation history, memory and skills carry over. Fallback providers are opt-in with hermes fallback, and Hermes Agent never switches models on a Nous Portal plan limit by itself: it says what ran out and offers free models, waiting for the reset, or credits. The status line shows the model in use.

Manus picks the model for each task by default. It offers its own modes (Manus 2.0 Lite, Manus 2.0 and Manus 2.0 Max), and free daily refresh credits work only in Lite. Manus Flex, launched in September 2026, lets you power Manus with your own API key from a supported provider; the first partners are OpenRouter, Fireworks and Modal. The provider bills inference directly, and other services in each task still use Manus credits.

It answers on more than 20 messaging platforms

Through one gateway, Hermes Agent connects to more than 20 messaging platforms, including Telegram, Discord, Slack, WhatsApp, Signal, iMessage through BlueBubbles, Microsoft Teams, LINE and email. Scheduled jobs can post their results to those chats, and in Telegram and Discord it can send spoken replies. One gateway owns every session, so the CLI, TUI, Hermes Desktop, bots and cron share one conversation.

A Hermes Agent mobile app is in beta and coming soon. Today you message it from your phone through a chat app, or use Hermes Desktop in your phone's browser with the Webapp.

Manus Agents work in Telegram on every plan, and in LINE and Slack. You can also start a task by forwarding an email to your Mail Manus address. Manus's separate Cue app, which currently requires an invitation code, supports iMessage, Telegram, LINE and Slack.

Your data is stored on the machine that runs it

Hermes Agent stores conversations, memory and skills in the Hermes Agent home directory on the machine that runs it (~/.hermes on macOS and Linux). Model requests go only to the providers you configure. With local models, model requests stay on your computer. Hermes Agent collects no telemetry by default; usage-metrics sharing is opt-in, and those metrics leave out prompts and responses.

Manus stores data in the US and Singapore. Its privacy policy covers files, shell commands and their outputs, code and task logs from its sandbox.

It remembers your work and writes its own skills

Hermes Agent maintains two curated memory files, a profile of your preferences and its own notes on your environment and what it has learned, and loads them at the start of every session. It stores every terminal and chat session and can search them all with full-text search.

When it works out a non-trivial workflow, it saves the approach as a skill and refines it with use. You can require review before a skill change lands, and a background curator archives skills that go unused.

Manus carries context through Projects, which apply a master instruction and a knowledge base of files to every new task in a project. Its user knowledge base holds up to 100 entries on Pro and 50 on Free, and you can ask Manus to save a successful task as a Skill.

Sources (33), checked November 9, 2026

Every statement about Manus on this page comes from the pages below. If something is out of date, tell us at support@nousresearch.com and we will correct it. Statements about Hermes Agent come from its documentation, source code and Nous Portal.

The Internet's Own AI

© 2026, Nous Research, Inc.

Terms|Privacy

MIT License · 2026