Hermes Agent v0.22 is the latest major release of Nous Research's free, open-source AI agent, and in this post I'll explain what it is and how a business, an agency or a creator can actually use it.
Nous Research calls it "The Threshold Release", and the official release notes date it 13 October 2026.
A patch called v0.22.1 followed on 14 October 2026, so that's the version you'll download today.
Here's the short version of what changed.
A brand-new desktop install now sets itself up through a chat, Hermes can work inside apps such as Linear and Notion, and you can run a model on your own hardware with one click.
On top of that, there's a 600-plugin catalog, a new wave of Claude and GPT-6 models, password-manager sign-in for the browser and a bot screen you can take over.
In my video I summed it up as three moves: walk in, plug in and hand off.
I'll explain each feature first, and then I'll show you the seven business jobs I'd give it.
This is a walkthrough of the official release notes, so I'll be clear about what Nous Research says each feature does rather than claim I've tested every one.
What Hermes Agent v0.22 Is, In Plain English
Hermes Agent is an AI worker that lives on your computer or server.
It uses tools, remembers what it learns, runs scheduled jobs and talks to you through apps such as Telegram, Slack and Discord.
Nous Research makes it, and it's free and open source under the MIT licence.
Version 0.22 is the largest release Hermes has shipped, at roughly 2.5 times the size of the v0.21 cycle according to the release notes.
The notes count about 24,200 commits and about 2,520 contributors since v0.21.0.
The odd bit is that the codebase got about a third smaller at the same time, dropping from roughly 1.06 million lines to about 698,000.
For a business owner, that matters because a leaner tool is usually easier to keep running.
The Problem This Release Solves For A Business
Before v0.22, getting Hermes running took real effort.
You needed an API key before you could even say hello, and a new team member faced a settings screen full of terms they didn't know.
Connecting it to the tools your team uses meant fiddling with MCP servers.
Running several profiles for different clients meant several processes on one machine, which could clash.
Updates could occasionally leave an install half-applied.
v0.22 goes after every one of those pain points, and that's why I think it's a big deal for people who aren't developers.
🔥 Want my exact What Is Hermes Agent v0.22 setup, step by step? Inside the AI Profit Boardroom you get step-by-step video trainings, my installable Agent OS and four coaching calls a week with 3,400+ members building real automations. → Join the AI Profit Boardroom
What's New In Hermes Agent v0.22, Mapped To Business Jobs
This table pairs each headline feature with the job it helps a business do.
| Feature | What Nous Research says it does | The business job it helps with |
|---|---|---|
| First-launch setup chat | A fresh desktop install sets itself up through a chat on a free trial model. | Getting a non-technical team member started without a setup call. |
| Connectors | Hermes acts inside hosted apps such as Linear and Notion after one browser sign-in. | Updating tasks and docs where your team already works. |
| One-click local models | Hermes installs llama.cpp and picks a model that fits your hardware. | Keeping sensitive drafts on your own machine. |
| 600-plugin catalog | Reviewed plugins install from the desktop app or the CLI. | Adding memory, search, voice or platform tools without custom code. |
| Password-manager sign-in | The browser agent logs in without the model seeing the password. | Letting the agent work in logged-in web tools more safely. |
| Bot Screen | A Linux-gateway bot gets a desktop you can watch and take over. | Stepping in for a login or 2FA, then handing the job back. |
| One gateway for every profile | One process per machine serves all your profiles. | Running separate bots for separate brands or clients. |
| Reasoning replay and outage waits | Long jobs keep their train of thought and wait out short provider errors. | Overnight and multi-step jobs that don't die halfway. |
| Stable releases and signed packages | Installs follow tested releases and the desktop app updates itself. | Fewer broken mornings after an update. |
How To Get Hermes Agent v0.22 Running
You don't need to be technical to get this release.
- If you use the desktop app, it updates itself, so accept the update when it appears.
- If you installed from the Microsoft Store, the Store updates it for you.
- If you use the command line, run
hermes update. - If you're starting fresh on Linux, macOS or WSL2, run
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash. - If you're starting fresh on native Windows, run
iex (irm https://hermes-agent.nousresearch.com/install.ps1)in PowerShell. - If you use Docker, pull
nousresearch/hermes-agent:stable.
Two requirements catch people out.
The new Windows desktop package needs Windows 11 22H2 or newer, and the default macOS build is for Apple Silicon only.
Use Case 1: Onboard A New Team Member Without A Setup Call
The first-launch setup chat is the feature I'd show a non-technical assistant first.
On a fresh desktop install, Hermes asks for a name, an accent colour and a layout, one card at a time.
Then it asks which apps to connect and which plugins to install, gives a quick tour and suggests up to four first tasks.
It runs on a free trial model called nous/welcome, once per operating-system user, so nobody needs an API key on day one.
If someone wants to redo it, they go to Settings → Run onboarding again.
The desktop app also has a new Simple mode that hides the developer panes and turns Hermes into a plain chat window.
Every user-facing string now runs through one translation catalog with full coverage in 16 bundled languages, which helps if your team works in more than one language.
The catch is that the free trial model is desktop-only, so the CLI, the TUI and the messaging gateway don't get it.
Use Case 2: Let Hermes Work Inside Your Project Tools
Connectors are where Hermes stops being a chat window and starts doing work in your stack.
Hermes can now act in hosted apps such as Linear and Notion through the Nous tool gateway.
You ask for the app, sign in once in your browser, and the agent can use it in the same turn.
You don't need a Nous account, because Hermes creates a free connectors-only identity the first time.
The part I like for agencies is named accounts.
You can connect several accounts for one app, such as a work account and a home account, and manage them with /connectors or hermes connectors.
A practical first job is asking Hermes to turn a meeting summary into tasks on your project board.
🔥 Want to build these Hermes workflows with help? Inside the AI Profit Boardroom, I've got Hermes Agent training, the Agent OS and step-by-step videos on connecting agents to real business tools. You'll be learning alongside 3,400+ members with four coaching calls every week. → Get access here
Use Case 3: Keep Sensitive Work On Your Own Hardware
Some work shouldn't leave your machine, such as private client notes or drafts under an NDA.
v0.22 adds one-click local models with no account and no API key.
Hermes installs llama.cpp with the right engine for your hardware and rates every catalog model against your GPU and RAM.
Then it picks the highest-quality build that fits, and it only recommends models that fit fully in GPU memory.
You start from Settings → Providers → Local Models in the desktop app, or hermes model then Local models in a terminal.
The catalog includes Qwen3.8 27B, Qwen3.8 Flash Next, Qwen3.6 35B-A3B and DeepSeek V4 Flash, plus any GGUF model from Hugging Face.
Be realistic, though, because a model that fits on a laptop won't do the same quality of work as a top hosted model.
I'd use local models for summaries, sorting and first drafts, and keep a stronger model for the work a client sees.
Use Case 4: Build A Toolkit From The Plugin Catalog
The plugin catalog launched this cycle and reached 600 entries.
Every plugin is pinned to a reviewed commit and installs from Hermes Desktop or the CLI.
The release notes break the new listings down like this.
| Catalog shelf | Number of listings |
|---|---|
| Tools | 180 |
| Desktop panes | 140 |
| Memory | 65 |
| Model providers | 48 |
| Web and search | 34 |
| Platforms | 32 |
| Voice | 18 |
The memory providers Honcho, Supermemory, Hindsight, Mem0, OpenViking and Holographic now live in the catalog with their makers.
Home Assistant, Spotify and Langfuse moved there too, and existing users are migrated automatically.
For a content business, the web and search shelf and the memory shelf are where I'd look first.
My Hermes Agent plugins marketplace guide explains how the catalog started, if you want the history.
Use Case 5: Let The Browser Agent Log In More Safely
A lot of business work sits behind a login, such as analytics dashboards and supplier portals.
The browser agent can now sign in with logins from 1Password, Bitwarden or Hermes's own encrypted vault.
The model never sees the password.
If a login isn't saved, you get a masked prompt to save it, and one-time codes are filled without entering the conversation.
Card and address fills at checkout need your confirmation.
My rule is simple: give the agent read-only accounts wherever the tool allows it, and never let it pay for anything on its own.
Use Case 6: Run Separate Bots For Separate Brands
If you run several brands or several clients, profiles are how you keep them apart.
Hermes now runs one messaging gateway and one desktop backend per machine for all your profiles.
Creating a profile makes its bots live without a restart.
Each profile keeps its own credentials, allowlists, cron jobs and sessions, so one client's bot doesn't see another client's keys.
Then there's Bot Screen.
A bot on a Linux gateway can have its own desktop, streamed live into Hermes Desktop.
When it hits a login, a 2FA prompt or a CAPTCHA, you take the screen, finish that step and hand control back.
With a Docker, SSH or Singularity backend, that screen and the browser run inside the sandbox rather than on your own computer.
Bot Screen is off by default and needs some host packages, so treat it as a setup job for an afternoon.
My Hermes Agent bot mode guide covers the bot side in more depth.
Use Case 7: Long Jobs That Survive A Bad Provider Hour
Long research jobs and overnight runs used to die when a provider had a wobble.
v0.22 replays prior reasoning by default on every provider that can read it, across resumes, branches and compaction.
When a provider throws a short burst of 503, 529 or timeout errors, Hermes waits it out in visible, bounded cycles instead of ending your turn.
Cron got sturdier as well.
A scheduled run that never reached the model, for example after your computer woke up, re-runs after 5, 15 and 30 minutes.
Jobs that aren't pinned now use whatever main model you've set at the moment they fire.
And you can now watch running subagents live, with Steer and Stop buttons, so you can correct one before it wastes time.
Choosing A Model For Business Work In v0.22
The release adds what Nous Research calls the Claude 5.5 and GPT-6 model wave.
That means Claude Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 5.5, plus GPT-6 Astra, Sol and Luna, GPT-6.1 Sol and Gemini 4 Argon.
They're pickable on Nous Portal, OpenRouter and native routes where the provider offers them.
The model picker now shows when a subscription is running low and when a rate limit lifts, so you can switch before a job stalls.
You can also set reasoning effort when you pick a model, which helps you save money on simple jobs.
For my own picks by job, see my ranking of the best Hermes Agent model.
The Limits A Business Should Know About
I'd rather you hear the limits from me than discover them on a client job.
- The free
nous/welcomemodel is a one-time desktop trial, not a free model for running your business. - Ended sessions inactive for 90 days are now deleted at startup unless you set
sessions.auto_prune: false. - Chats no longer reset on a timer, so you need
/newor/resetto start fresh. - The Windows desktop app needs Windows 11 22H2 or newer, and the default Mac build is Apple Silicon only.
- Bot Screen needs a Linux gateway and some host packages.
- Local models trade quality for privacy, so test them on real work before you trust them.
What The v0.22.1 Patch Changed
v0.22.1 arrived on 14 October 2026.
Nous Research describes it as a patch with a handful of fixes for the Windows Store, plugin updates and more.
It also adds a curated model list at the top of the CLI model picker.
New catalog plugins include Korean and Brazilian Portuguese language packs for the desktop app.
🔥 Want the Agent OS with Hermes profiles ready to go? The AI Profit Boardroom has the installable Agent OS, Hermes training and four weekly coaching calls where you can ask me which business jobs to hand to which agent. → Join the AI Profit Boardroom
The Honest Verdict For Business Use
v0.22 is the first Hermes release I'd happily hand to a non-technical team member.
The setup chat removes the scary first screen, connectors put the agent where your team already works, and local models give you a private option.
The single gateway makes running bots for several clients much tidier.
The weak spots are the hardware requirements and the fact that the free model is only a trial.
Start with one job, such as turning meeting notes into tasks, and add the next feature only when that first job runs cleanly.
Related Reading
- Hermes Agent plugins marketplace, explained
- Hermes Agent bot mode for business
- What an AI SEO optimisation agency actually does
Also On Our Network
- 🌐 The step-by-step Hermes Agent v0.22 update and setup
- 🌐 Hermes Agent v0.22 feature by feature, with scores
- 🌐 How Hermes Agent v0.22 fits into an Agent OS
- 🌐 Benchmark numbers for the models behind v0.22
FAQ: Hermes Agent v0.22 For Business
What is Hermes Agent v0.22 in simple terms?
It's a big update to Hermes Agent, the free and open-source AI agent from Nous Research.
It makes Hermes easier to start, lets it work inside your apps and runs models on your own hardware in one click.
Can a business use Hermes Agent v0.22 for free?
Hermes Agent itself is free and MIT licensed.
You pay only for hosted models or subscriptions if you choose them.
Which apps can Hermes Agent v0.22 connect to?
The release notes name hosted apps such as Linear and Notion through the Nous tool gateway.
The Connectors page lists those apps and your MCP servers together.
Is Hermes Agent v0.22 safe to use with client logins?
The browser agent signs in through 1Password, Bitwarden or Hermes's own vault, and the model never sees the password.
You should still start with low-risk accounts.
How do I get Hermes Agent v0.22?
The desktop app updates itself, and a command-line install updates with hermes update.
If you run an agency and want to talk through where AI agents fit in your SEO work, you can book a free SEO strategy session with my team.
📺 Video notes + links to the tools 👉
🎥 Learn how I make these videos 👉
🆓 Get a FREE AI Course + Community + 1,000 AI Agents 👉
About Julian
I'm Julian Goldie, an SEO entrepreneur, author and founder of the AI Profit Boardroom, which has 3,400+ members.
I help business owners scale with AI agents, automation and SEO.
- I've built a 7-figure SEO and link-building agency, Goldie Agency, with 70+ people in Bangkok.
- I've grown a YouTube channel to 400,000+ subscribers.
- I wrote the book "Link Building Mastery", which covers nine link building strategies.
- I run GoldieBench, where I score AI models on real one-shot builds.
→ Get my best AI training inside the AI Profit Boardroom
Pick one business job, switch on the matching feature and build from there, and you'll get real value out of Hermes Agent v0.22.
