agentos.guide › blog
Hermes MissingNo is the new free stealth model you can run inside Hermes Agent, and Nous Research has made it free for a limited time on Nous Portal.
It is a coding and agentic reasoning model, and nobody has said which lab built it.
You can switch it on in Hermes Cloud in about ten clicks, or run it as its own Hermes profile.
I tested it, and it was great at the /learn command, fast at tool calls and only moderate at coding.
My MissingNo video is not public yet, so the video below is my Solar Mini 4 walkthrough, which shows the same Hermes Cloud free-model screens.
Hermes MissingNo means running the MissingNo model as the brain of a Hermes Agent.
MissingNo is a stealth model, which means you can use it but the lab that built it is hidden for now.
The name is a Pokémon reference, because MissingNo is short for "missing number", the glitch Pokémon from the first games.
Nous Research, the makers of Hermes Agent, announced it on X on 10 October 2026.
Their post opened with "Wild MISSINGNO. appeared!", so they are clearly enjoying the joke.
| Fact | What I could verify on 12 October 2026 |
|---|---|
| Who released it | Nous Research made it available on Nous Portal and announced it on X on 10 October 2026. |
| Who built it | The lab has not been announced, which is why it is a stealth model. |
| What it is for | Nous describes it as a coding and agentic reasoning model that is highly token efficient. |
| Cost | It is free for a limited time, and no end date has been published. |
| Model ID | The Nous inference API lists it as stealth/missingno. |
| Context window | The listing shows 262,144 tokens, with up to 131,072 output tokens. |
| Inputs | It takes text only. |
| Tools | The listing shows tool calling, parallel tool calls and structured outputs. |
| Reasoning | Reasoning is always on, with low, medium, high, xhigh and max levels, and medium is the default. |
| Not published | I could not find the maker, a parameter count or an official benchmark. |
I have guessed out loud that it could be Nous Research's own model.
That is only a guess, and nothing official supports it.
If you want the company background, read my Hermes Agent by Nous Research guide.
MissingNo is free right now.
The Nous inference API showed a price of zero for input and output when I checked.
The catch is in the words "limited time".
Nous Research did not say when the free window ends.
A stealth model can be removed, renamed or turned into a paid model with very little notice.
So I treat it as a free bonus, and I never make it the only brain a workflow can run on.
Hermes Cloud is the fastest way in, because the agent runs on Nous Research's servers.
| Step | What you click in Hermes Cloud |
|---|---|
| 1 | Open portal.nousresearch.com/cloud and click Deploy. |
| 2 | Log in to Nous Portal. |
| 3 | Click Create agent. |
| 4 | Open Manage agents. |
| 5 | Click on the agent you just created. |
| 6 | Click Open dashboard. |
| 7 | Go to the model section. |
| 8 | Search "Nous Research", or type "missing". |
| 9 | Pick the stealth model, which is MissingNo. |
| 10 | Reload the page and start chatting. |
The reload in step ten matters, because the chat can hold on to the old model until the page refreshes.
There is a shortcut as well.
You can ask Hermes to configure Hermes Cloud for you, and the agent will change its own model.
I still open the model section afterwards to confirm it says MissingNo.
When I checked the Nous inference API, I saw one MissingNo entry and it was priced at zero.
That means I did not see a paid twin next to it, which is the trap I warned about in my Solar Mini 4 Hermes guide.
🔥 Want the exact free Hermes stealth model setup I run inside my Agent OS?
Inside the AI Profit Boardroom you get the installable Agent OS, step-by-step video tutorials, four coaching calls a week and 3,400+ members building real automations.
I keep one Hermes profile per model inside my Agent OS.
A profile is an isolated Hermes with its own model, settings and memory.
Right now I have a Step 5 Preview profile, which is another free model, and a MissingNo profile beside it.
That lets me test a new brain without touching the agent I use every day.
On your own machine, hermes portal logs you in to Nous Portal.
The hermes profile create missingno command makes the profile.
The hermes dashboard command opens the web dashboard, where Models lets you set the main model.
Running hermes -p missingno opens a chat in that profile.
If the model is missing from the picker, hermes model --refresh re-fetches every provider's live model list.
In the Agent OS, I open the Hermes tab, pick the missingno profile and start chatting.
I do not trust a new brain until I have run real jobs through it.
| Test | What happened | My read |
|---|---|---|
/learn | It created a skill really well and fast, and it named the path. | This was the best result, and it is the job I would give it. |
| Coding | A simple ping-pong style game took about five minutes, and it works. | It is moderate and it is not frontier level. |
| Tool calling | It was fast at calling agentic tools. | It suits agent loops more than large builds. |
| Reply speed | Replies came back quickly on Hermes Cloud. | Hosted models respond fast. |
The /learn command points Hermes at a source and turns it into a reusable skill.
MissingNo handled that really well, and it was fast.
It created the skill and told me the path where the skill lives.
The coding task did not work so well.
I asked for a simple ping-pong style game, and it took around five minutes to generate.
The game works, but it is not frontier level, and it will not replace Claude Opus 5.5.
It was quick at calling agentic tools, which is what an agent spends most of its day doing.
Hermes Cloud models also respond quickly in general, because they are hosted for you.
Most people treat a new model like a new chatbot, and I think that is a mistake.
In my Agent OS, a model is a part I slot into a profile.
The workflow lives in Hermes, in the skills, the tools and the memory.
The model is one setting, and I can change it in seconds.
That is what makes a stealth model safe to play with.
If MissingNo vanishes tomorrow, the skills it built with /learn are still on disk.
I just point the profile at a different brain and carry on.
| Job in the Agent OS | Brain I would give it | Why |
|---|---|---|
Turning documents into skills with /learn | MissingNo | It did this really well and fast in my test. |
| Competitor analysis in Hermes Oracle | MissingNo, with a human check | Research is mostly tool calls, and it calls tools quickly. |
| A trial Hermes Bot | MissingNo | A bot made from the active profile costs nothing to test. |
| Small scripts and fixes | MissingNo or another free model | Coding was moderate, so keep the stakes low. |
| Real builds and client code | A frontier model such as Claude Opus 5.5 | MissingNo will not replace it. |
| Anything that pays, sends or deletes | A human approval step | No model should own an action you cannot undo. |
I also showed a voice agent inside my Agent OS.
It is a pop-up voice agent that works with ChatGPT and Hermes.
I asked it for useful Hermes commands, and it listed things like /learn and /skills.
Those are the two commands I would try first on any new model.
The /learn command builds a skill, and /skills shows you what your agent already knows.
MissingNo is not limited to the chat window.
You can use it in Hermes Bots by creating a bot from your active profile.
You can use it in Hermes Oracle, where competitor analysis is the job I would start with.
You can use it in any other workflow that reads the model from that profile.
My Hermes Bot Mode guide covers the bot side, and the new Hermes Agent Bot Mode explainer goes through the feature step by step.
🔥 Want the exact free Hermes stealth model setup I run inside my Agent OS?
Inside the AI Profit Boardroom you get the installable Agent OS, step-by-step video tutorials, four coaching calls a week and 3,400+ members building real automations.
This is the step that keeps your agents running.
MissingNo might be taken down, it might become paid, or you might run out of tokens.
When that happens, open the model section and type "free".
There were about eight free models when I recorded.
| Free model | Maker | Role in my Agent OS |
|---|---|---|
| MissingNo | Unannounced | The stealth brain for skills and tool-heavy jobs. |
| Step 5 Preview | StepFun | The other free model I keep a profile for. |
| Step 3.7 Flash | StepFun | The quick everyday backup. |
| Ling 3.0 | inclusionAI | A second free flash option for cross-checking. |
| LongCat | Meituan | The one I reach for when the input is very long. |
| Laguna XS 2.1 | Poolside | A small model for light coding chores. |
| Solar Mini 4 | Upstage | A fast model, and the free entry is the one to pick. |
The free list changes often, so check it inside Hermes on the day you need it.
My free models in Hermes guide goes through the wider options.
Because I keep one profile per model, switching is a two-click job in the Hermes tab.
MissingNo takes text only, so describe images in words or use another model.
Reasoning is always on, so simple questions can take a little longer than you expect.
The context window is 262,144 tokens, which is plenty for most agent jobs.
There are no official benchmarks that I could find, so my tests are all I can give you.
One test is not a benchmark, and your jobs may go better or worse than mine did.
I would not put a stealth model in front of customers, because it can change overnight.
I would happily let it build skills, run research and power an internal bot.
/learn on a short document and note the path it gives you./skills and confirm the new skill appears.If it passes the first four, it has earned a place in your fast lane.
If the fifth is slow, you have seen the same thing I saw.
Deploy a Hermes Cloud agent, type "missing" in the model section and pick the stealth model.
Run the five-minute test plan before you give it anything important.
Then decide which backup model takes over when the free window closes.
If you want the whole Agent OS with the Hermes tab, the profiles and the training already set up, it lives inside the AI Profit Boardroom, with 3,400+ members and four coaching calls a week.
Yes, for now. Nous Research said MissingNo is free for a limited time on Nous Portal and gave no end date, so it may be removed or become a paid model later.
Click Deploy, log in, click Create agent, open Manage agents, click the agent, click Open dashboard, go to the model section, type missing, pick the stealth model, then reload and chat.
The lab has not been announced. Nous Research made it available on Nous Portal as a stealth model. Julian has guessed it could be Nous's own model, and that is only a guess.
The Nous inference API lists it as stealth/missingno with a 262,144-token context window, text input only, tool calling and always-on reasoning.
It is moderate. In Julian's test a simple ping-pong style game took about five minutes to generate and it worked, but it was not frontier level and it will not replace Claude Opus 5.5.
Type free in the model section and pick another free model, such as Step 3.7 Flash, Step 5 Preview, Ling 3.0, LongCat, Laguna XS 2.1 or Solar Mini 4.
Make a separate profile, give it skills and tool-heavy jobs, and keep a backup model ready for the day the free window closes.
That is how you get real value out of Hermes MissingNo while it lasts.
About Julian
I'm Julian Goldie, an AI entrepreneur, SEO expert and the founder of the AI Profit Boardroom, which has 3,400+ members.
I help business owners scale with AI agents, automation and SEO.
My YouTube channel has 400,000+ subscribers, and I run Goldie Agency, a seven-figure SEO agency.
→ Solar Mini 4 Hermes: The Quick Start
→ Hermes /learn: The Book Brain Engine
The installable system behind every guide on this site — dashboard, agents, memory and pipelines — updated as new tools land, with 3,400+ members and four coaching calls a week.
Join the AI Profit Boardroom →📺 Video notes + links to the tools 👉
🎥 Learn how I make these videos 👉