Hermes Agent v0.22 is the newest major release of the free, open-source AI agent from Nous Research, and this page explains every headline feature, scores each one and ends with my verdict.
Nous Research named it "The Threshold Release", and its official release notes carry the date 13 October 2026.
A patch, v0.22.1, followed on 14 October 2026.
A fresh desktop install now opens on a setup chat that runs on a free trial model, and Hermes can now work inside your apps, run local models in one click and update without breaking.
In my video I described it as walk in, plug in and hand off, and that's still the simplest way to hold it in your head.
One thing to be clear about up front is that my scores come from reading the official release notes and from running Hermes in my own Agent OS, not from a lab test of each new feature.
Hermes Agent v0.22 Verdict At A Glance
These are my scores out of 10 for how much each feature changes daily use, based on the release notes and my own Hermes setup.
| Feature | What it does | My score |
|---|---|---|
| First-launch setup chat | It sets up a fresh desktop install through a chat on a free trial model. | 9 |
| Connectors | It lets Hermes act inside hosted apps such as Linear and Notion. | 9 |
| One-click local models | It installs llama.cpp and picks a model that fits your hardware. | 8 |
| Stable release channels | It moves installs onto tested SemVer releases by default. | 8 |
| Signed packages and one package manager | It makes the desktop app self-updating and stops half-applied updates. | 8 |
| 600-plugin catalog | It lists reviewed plugins you can install from the app or CLI. | 8 |
| Claude 5.5 and GPT-6 model wave | It adds the newest Claude, GPT-6 and Gemini 4 models. | 7 |
| Password-manager sign-in | It lets the browser agent log in without the model seeing passwords. | 8 |
| Bot Screen | It gives a Linux-gateway bot a desktop you can take over. | 7 |
| One backend and one gateway | It serves every profile from one process per machine. | 8 |
| Live subagents | It shows each subagent's work with Steer and Stop buttons. | 8 |
| Smarter model picker | It shows quota, rate-limit resets and reasoning effort. | 7 |
| Leaner codebase and prompts | It cuts about a third of the code and trims the prompt. | 7 |
My overall rating is 4.5 out of 5, and I explain the half point I held back in the limits section.
๐ฅ Want my exact Hermes Agent v0.22 Explained 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. โ Join the AI Profit Boardroom
What Hermes Agent v0.22 Is
Hermes Agent is an open-source AI agent that runs on your own machine or server, uses tools, learns skills and talks to you through chat apps.
These are the release facts I checked against the official notes, and my Hermes Agent by Nous Research guide covers the team behind them.
| Item | Detail |
|---|---|
| Name | Hermes Agent v0.22.0, "The Threshold Release". |
| Maker | Nous Research builds and maintains Hermes Agent. |
| Release date | The release notes are dated 13 October 2026. |
| Patch | v0.22.1 followed on 14 October 2026. |
| Licence and cost | It's free and open source under the MIT licence. |
| Size of the cycle | The notes count about 24,200 commits, 6,080 merged PRs, 7,180 closed issues and 2,520 contributors since v0.21.0. |
| Scale versus last cycle | Nous says it's about 2.5 times the size of the v0.21.0 cycle. |
| Roll-up | It documents the v0.21.1 to v0.21.7 patch releases too. |
How To Get Hermes Agent v0.22
The desktop app updates itself, and Microsoft Store installs update through the Store.
A command-line install updates with hermes update.
A fresh install on Linux, macOS or WSL2 uses curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash, and native Windows uses iex (irm https://hermes-agent.nousresearch.com/install.ps1) in PowerShell.
Docker users pull nousresearch/hermes-agent:stable.
Feature 1: The First-Launch Setup Chat
A fresh desktop install now starts a short setup chat instead of a settings screen.
It asks for your name, an accent colour and a layout, then which apps to connect and which plugins to install.
It finishes with a quick tour and up to four first tasks picked for you.
The chat runs on the free nous/welcome trial model, once per operating-system user, so it needs no key.
Skip setup ends it, and Settings โ Run onboarding again brings it back.
The agent also gained a set_theme tool, so "make it violet, dark mode" works in any desktop chat.
My verdict: this is the single biggest win for beginners, and I score it 9 out of 10.
It loses a point because existing installs, remote backends and the CLI don't get it.
Feature 2: Connectors
Hermes can now act in hosted apps such as Linear and Notion through the Nous tool gateway.
You sign in once in your browser, and the tools work in the same turn.
You don't need a Nous account to start, because Hermes makes a free connectors-only identity the first time.
A new Connectors page puts hosted apps and MCP servers in one list, and it replaces the MCP tab.
You can connect several named accounts for one app and manage them with /connectors or hermes connectors.
My verdict: this scores 9 out of 10, because it's what turns Hermes from a chat into a worker inside your tools.
Feature 3: One-Click Local Models
Hermes now installs and maintains llama.cpp with the right engine for CPU, CUDA, Metal, Vulkan or ROCm.
It rates every catalog model against your GPU and RAM and picks the highest-quality build that fits.
You start from Settings โ Providers โ Local Models or hermes model then Local models.
The catalog includes Qwen3.8 27B, Qwen3.8 Flash Next, Qwen3.6 35B-A3B and DeepSeek V4 Flash, plus any Hugging Face GGUF.
My verdict: this scores 8 out of 10, because it removes the hardest part of running local models, which is picking the right one.
The two points I held back are for the obvious reason that local models are still weaker than the best hosted ones.
Feature 4: Stable Release Channels
Hermes now ships tested SemVer releases.
Each release passes source CI, Docker, Nix, signed desktop builds and install and upgrade runs on every operating system before it's published.
Source installs now move from release to release by default instead of tracking every commit on main, and hermes update --set-channel main switches you back to the bleeding edge.
My verdict: this scores 8 out of 10, because boring updates are exactly what a working setup needs.
Feature 5: Signed Desktop Packages And One Package Manager
The desktop app is now a sealed, signed package that carries its own Python and Node and updates itself.
Windows gets a signed MSIX for x64 and ARM64 through App Installer or the Microsoft Store, and macOS gets a signed, notarised Apple Silicon build.
Underneath sits a new package manager, hermes pm, which owns every pinned, hash-checked tool and Python environment.
Updates build a new environment and switch only once it passes a check, so a killed update can't leave you half-applied.
My verdict: this scores 8 out of 10.
The requirement for Windows 11 22H2 or newer and Apple Silicon on the default Mac build is the catch.
๐ฅ Want my full Hermes setup and model-testing workflow? Inside the AI Profit Boardroom, I've got a Hermes Agent course, the installable Agent OS and step-by-step videos on every big Hermes release. You'll be learning alongside 3,400+ members with four coaching calls every week. โ Get access here
Feature 6: The 600-Plugin Catalog
The plugin catalog launched this cycle and grew to 600 entries, all new in this release window.
Each listing is pinned to a reviewed commit, and you can install it from Hermes Desktop or the CLI.
You can browse by category and star ranking, with a page for each plugin and each author on the Hermes plugin docs.
The release notes count 180 tools, 140 desktop panes, 65 memory plugins, 48 model providers, 34 web and search plugins, 32 platforms and 18 voice plugins.
Memory providers such as Honcho, Supermemory, Hindsight, Mem0, OpenViking and Holographic now live in the catalog with their makers.
My verdict: this scores 8 out of 10, and I like that the rules are strict.
For my own shortlist, see the best Hermes Agent plugins.
Feature 7: The Claude 5.5 And GPT-6 Model Wave
The release notes list Claude Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 5.5.
They also list GPT-6 Astra, Sol and Luna, GPT-6.1 Sol and Gemini 4 Argon.
These are pickable on Nous Portal, OpenRouter and native routes where the provider offers them.
The catalog also adds models such as DeepSeek V4.1 Flash, Grok 4.7, GLM-5.3-FlashX, Muse Spark 1.3 and Solar Pro 4 and Mini 4.
Prior reasoning is now replayed by default on every provider that can read it, across resume, branches and compaction.
When a provider has a bad hour, Hermes waits out short 503, 529 and timeout errors in visible, bounded cycles instead of ending your turn.
My verdict: this scores 7 out of 10, because new model names matter less than the reasoning replay and outage handling that ship with them.
Feature 8: Password-Manager Sign-In
The browser agent can now log into sites with 1Password, Bitwarden or Hermes's own encrypted vault.
The model never sees the password.
One-time codes are filled without entering the conversation, and card and address fills need your confirmation.
My verdict: this scores 8 out of 10, because it fixes the worst part of browser agents, which is pasting secrets into a chat.
Feature 9: Bot Screen
A bot running on a Linux gateway can now have its own desktop, streamed live into Hermes Desktop.
When it hits a login, a 2FA prompt or a CAPTCHA, you take over the screen, finish that step and hand it back.
With a Docker, SSH or Singularity backend, the screen, computer use and browser run inside the sandbox rather than on your host.
It's off by default and needs host packages.
My verdict: this scores 7 out of 10, because it's powerful but only for people already running bots on Linux.
Feature 10: One Backend And One Gateway For Every Profile
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, and each profile keeps its own credentials, allowlists, cron jobs and sessions.
hermes update migrates per-profile setups automatically when nothing conflicts.
My verdict: this scores 8 out of 10, because fewer processes and no token clashes make multi-profile setups far calmer.
Feature 11: Live Subagents You Can Steer
Running subagents now stay visible above the composer with each child's task, elapsed time and live transcript.
You open the view with F6 in the CLI or Ctrl+T in the TUI, and it's also in the desktop app.
Each child gets Steer and Stop buttons, and /stop now hands back a background child's partial work.
My verdict: this scores 8 out of 10, because seeing what a subagent is doing is how you stop wasted runs.
Feature 12: A Model Picker That Knows Your Plans
You can star favourite models and see a chip such as "17% left ยท resets 3:00 PM" when a subscription is running low.
The picker shows when a rate limit lifts, and you can set reasoning effort as you choose the model.
My verdict: this scores 7 out of 10, because it's a quality-of-life change rather than a new capability.
Feature 13: A Leaner Codebase And Lighter Prompts
The source went from about 1.06 million lines to about 698,000, a cut of 34%, with zero behaviour change according to Nous.
About 19 rarely used tools now load only when the agent searches for them, which roughly halves the desktop tool schema sent on every call.
Another 2,100 or so tokens came off the fixed prompt.
My verdict: this scores 7 out of 10, because you'll feel it as slightly cheaper and faster calls rather than as a new button.
Everything Else In Hermes Agent v0.22
The notes are long, so here are the smaller changes worth knowing.
- All user-facing text now runs through one translation catalog with full coverage in 16 bundled languages.
- Ctrl+C clears your draft instead of killing the turn, and
/queueis now editable. - The desktop app has a Simple mode that hides developer panes.
- Unpinned cron jobs now use your main model at the moment they fire.
- A cron run that never reached the model re-runs after 5, 15 and 30 minutes.
- A stalled-turn watchdog interrupts a turn that makes no progress for 600 seconds by default.
What Was Reverted And Isn't Shipping
- Collective Wisdom V1, an in-tree skill marketplace, was merged and then removed.
- Approval and clarify prompts that wait until you answer were reverted, so
approvals.timeoutstill applies. - Parallel group-chat rounds in Bot Mode were reverted, so rooms still run one at a time.
The Hermes Agent v0.22.1 Patch
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 adds a curated model list at the top of the CLI model picker and new catalog plugins, including Korean and Brazilian Portuguese language packs.
The Limits, And Why It's Not A Perfect Score
This is where the missing half point went.
- The free
nous/welcomemodel is a one-time desktop trial, and the CLI, TUI and gateway don't get a free model. - Ended sessions inactive for 90 days are now deleted at startup unless you set
sessions.auto_prune: false. - The Windows desktop app is MSIX-only and needs Windows 11 22H2 or newer.
- The default macOS build supports Apple Silicon only.
- Python 3.14 is the managed runtime, and KittenTTS doesn't work on it.
- The
hermes synccommand and the bundled tirith scanner were removed.
Who Should Update To Hermes Agent v0.22
Beginners should install it, because the setup chat is the easiest way into Hermes so far.
Existing users should update, but they should set the auto-prune option first if they keep old chats.
People on Intel Macs or Windows 10 should use the command-line install rather than the new desktop package.
๐ฅ Want to ask me which Hermes features to switch on first? The AI Profit Boardroom has Hermes training, the Agent OS and four weekly coaching calls where you can bring your setup and get a straight answer. โ Join the AI Profit Boardroom
Final Verdict On Hermes Agent v0.22
I rate it 4.5 out of 5.
The setup chat and connectors are the two features that change who can use Hermes, and the stable channels and signed packages make it safer to keep running.
If you only switch on one thing, switch on connectors, because that's where Hermes starts doing work in your tools.
Related Reading
- The best Hermes Agent plugins
- Hermes Agent by Nous Research, explained
- My verdict on choosing an AI SEO optimisation agency
Also On Our Network
- ๐ Update to Hermes Agent v0.22 and switch on each feature
- ๐ Seven business uses for Hermes Agent v0.22
- ๐ A Hermes Agent v0.22 quick start inside an Agent OS
- ๐ GoldieBench scores for the models you can run in v0.22
FAQ: Hermes Agent v0.22 Review
Is Hermes Agent v0.22 worth updating to?
For most people it is, because it adds the setup chat, connectors, local models, plugins and Bot Screen.
Read the upgrade notes first, because old sessions are now pruned after 90 days.
What is the biggest new feature in Hermes Agent v0.22?
For beginners it's the first-launch setup chat, and for existing users it's connectors.
When was Hermes Agent v0.22 released?
The v0.22.0 release notes are dated 13 October 2026, and v0.22.1 followed on 14 October 2026.
Which models does Hermes Agent v0.22 add?
The notes list Claude Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 5.5, GPT-6 Astra, Sol and Luna, GPT-6.1 Sol and Gemini 4 Argon.
Did anything get removed in Hermes Agent v0.22?
Yes, hermes sync, the old plugin compatibility layer and the bundled tirith scanner were removed, and timed session resets were retired.
If you want a second opinion on where agents like Hermes could fit your SEO, you can book a free SEO strategy session with my team.
๐บ Video notes + links to the tools ๐
๐ฅ Learn how I make these videos ๐
๐ Get a FREE AI Course + Community + 1,000 AI Agents ๐
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.
โ Get my best AI training inside the AI Profit Boardroom
Switch on connectors first, read the limits, and you'll see why I rate Hermes Agent v0.22 so highly.











