Subject: Re: Can anyone get Edge Copilot to work without logging in after July 2026?

On Mon, 7/27/2026 9:20 PM, Maria Sophia wrote:

Can anyone get Edge Copilot to work without logging in after July 2026?
https://copilot.microsoft.com/

As of this month, whenever I use MS Edge to go to Copilot, no matter what tricks I use, it wants me to immediately log into a mothership account.

I can only find one trick that still works, and it only works in Edge.
But with no other browser can I get copilot to not ask for a login now.

Q: Other than sidebar Copilot, did you Copilot in Edge just recently ask
for a login to a mothership account every time you go to it using Edge?

NB: A separate thread on the Firefox ng shows nobody can do it there.
Newsgroups: alt.comp.software.firefox
Subject: Recently, did your Firefox stop using Microsoft Copilot in anonymous mode?
Date: Sun, 26 Jul 2026 14:00:07 -0700
Message-ID: <1145sgm$1cju$1@nnrp.usenet.blueworldhosting.com>

On my W11 Home (25H2), if I open MSEdge, there is a CoPilot icon right-most
on the top line. That is what I would use for CoPilot "Chat".

My Task Bar also has a CoPilot icon, but that icon "doesn't know who I am". That would be an example of a path with a login requirement. At first, it
made a fuss of installing itself, but now it sits there quietly.
I think there was an MSOFFice365 CoPilot6 that made a fuss of
installing itself, and I hunted that down and removed it (since
M365 isn't rented/used on the machine and who wants a grumpy AI
with nothing to play with).

Yet, the MSEdge and its top-right-most icon, it knows my username, which implies that perhaps my usage of a PIN at startup time, counts as
a login, and that's why I am not being pestered there. No policy change
has occurred there recently.

I don't fool with any other CoPilot portals, because if I were to do
that, there would undoubtedly be some "side effect" I would later
regret having triggered.

Paul

--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Can anyone get Edge Copilot to work without logging in after July 2026?

On 7/27/26 9:20 PM, Maria Sophia wrote:

Can anyone get Edge Copilot to work without logging in after July 2026?
https://copilot.microsoft.com/

As of this month, whenever I use MS Edge to go to Copilot, no matter what tricks I use, it wants me to immediately log into a mothership account.

I can only find one trick that still works, and it only works in Edge.
But with no other browser can I get copilot to not ask for a login now.

Q: Other than sidebar Copilot, did you Copilot in Edge just recently ask
for a login to a mothership account every time you go to it using Edge?

NB: A separate thread on the Firefox ng shows nobody can do it there.
Newsgroups: alt.comp.software.firefox
Subject: Recently, did your Firefox stop using Microsoft Copilot in anonymous mode?
Date: Sun, 26 Jul 2026 14:00:07 -0700
Message-ID: <1145sgm$1cju$1@nnrp.usenet.blueworldhosting.com>

In Linux, I have a bookmark that I launch.
Now the first time I ran it tonight I had to log in, but the next 4 times I loaded Firefox
and launched Copilot, I didn't have to log in. Time would only tell.

--
Mint 22.3, Thunderbird 153.0esr, Firefox 153.0
Alan K.

--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Can anyone get Edge Copilot to work without logging in after July 2026?

On Mon, 27 Jul 2026 22:10:08 -0400, "Alan K." <alan@invalid.com>
wrote:

On 7/27/26 9:20 PM, Maria Sophia wrote:

Can anyone get Edge Copilot to work without logging in after July 2026?
https://copilot.microsoft.com/

As of this month, whenever I use MS Edge to go to Copilot, no matter what
tricks I use, it wants me to immediately log into a mothership account.

I can only find one trick that still works, and it only works in Edge.
But with no other browser can I get copilot to not ask for a login now.

Q: Other than sidebar Copilot, did you Copilot in Edge just recently ask
for a login to a mothership account every time you go to it using Edge?

NB: A separate thread on the Firefox ng shows nobody can do it there.
Newsgroups: alt.comp.software.firefox
Subject: Recently, did your Firefox stop using Microsoft Copilot in anonymous mode?
Date: Sun, 26 Jul 2026 14:00:07 -0700
Message-ID: <1145sgm$1cju$1@nnrp.usenet.blueworldhosting.com>

In Linux, I have a bookmark that I launch.
Now the first time I ran it tonight I had to log in, but the next 4 times I loaded Firefox
and launched Copilot, I didn't have to log in. Time would only tell.

Could it be that your Firefox has cookies set that tell it that you
are *still* logged in from the first session? "Keep me logged in" is
an option I sometimes see on sites.

It might be fun to try to log out of Copilot the clear the cookies
then restart a Firefox session and see if it asks for a login.

J.


--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Can anyone get Edge Copilot to work without logging in after July 2026?

On Mon, 7/27/2026 9:20 PM, Maria Sophia wrote:

Can anyone get Edge Copilot to work without logging in after July 2026?
https://copilot.microsoft.com/

As of this month, whenever I use MS Edge to go to Copilot, no matter what tricks I use, it wants me to immediately log into a mothership account.

I can only find one trick that still works, and it only works in Edge.
But with no other browser can I get copilot to not ask for a login now.

Q: Other than sidebar Copilot, did you Copilot in Edge just recently ask
for a login to a mothership account every time you go to it using Edge?

NB: A separate thread on the Firefox ng shows nobody can do it there.
Newsgroups: alt.comp.software.firefox
Subject: Recently, did your Firefox stop using Microsoft Copilot in anonymous mode?
Date: Sun, 26 Jul 2026 14:00:07 -0700
Message-ID: <1145sgm$1cju$1@nnrp.usenet.blueworldhosting.com>

There is "registration" and there is "logging in".

CoPilot is listed as "no registration", but if it is
actually powered by an MSA login, then that amounts to registration.

If you don't have an MSA on the machine, then they can't tap
into the MSA scheme. That's probably why my CoPilot sessions
which are inside MSEdge, are friction free, as they know who
I am, and it is my entry of the PIN at the beginning of
a session which counts as an MSA login. You might not be able
to set up a PIN, without an MSA, just a guess on my part, as
I haven't tried to set up a PIN on anything other than the
Daily Driver (which happens to have an MSA).

https://inventwithpython.com/blog/free-llms.html

"Assume that free services may have daily usage limits, and that sites that
don't require registration may limit how much you can use them before
signing up for a free account.

You should absolutely assume that all content you submit as well as your
browser settings and IP address are recorded, sold, and matched to your
personal identity otherwise by the AI company.
"

"My rough top recommendations for coding tutors are ChatGPT, Claude, and DeepSeek,
but here are the acceptable free AI services I've found:

Duck.ai (from DuckDuckGo) - No registration required. Uses other AI companies' models
but acts as an anonymous relay. (This is not verified.)
ChatGPT (from OpenAI) - No registration required.
Gemini (from Google) - No registration required.
Claude (from Anthropic) - Free account registration required.
DeepSeek (from DeepSeek) - Free account registration required.
Copilot (from Microsoft) - No registration required.
Poe (from Twitch.tv) - Free account registration required.

I don't recommend the following for their deceptive UI practices
(They appear free but require registration after you've submitted a prompt):

Perplexity (from Perplexity)
Mistral (from Mistral AI)
Meta (from Meta / Facebook)
Grok (from X / Twitter)

I have slight recommendations against the following:

Shiori (from Shiori.ai) - Free account registration required. Uses other
AI companies' models. I don't recommend it because
this probably means their business model is personal
information collection. They may be fine, but they don't
have the reputation of DuckDuckGo for privacy.

(Updated June 12, 2026)
"

As far as I know, the CoPilot.App always looked for an MSA via a Microsoft login,
whereas the MSEdge version is likely obtaining your details via other means. You would think the CoPilot.App via WebView2, would be as adept as MSEdge
at this, but maybe not.

You can keep an AI model locally, but that's an acquired taste.
Models range from Google Gemma at 4GB, to the largest Kimi model at
1.6TB RAM and 100GB+ VRAM (four hundred times bigger). And with the
current RAM pricing, building a machine with that much RAM, you'll
need to win the lottery to be able to afford it. That's about 100K
worth of RAM, for that big model. The processor by comparison,
is "almost free" :-)

Paul

--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Can anyone get Edge Copilot to work without logging in after July 2026?

On 7/27/26 11:14 PM, John wrote:

On Mon, 27 Jul 2026 22:10:08 -0400, "Alan K." <alan@invalid.com>
wrote:

On 7/27/26 9:20 PM, Maria Sophia wrote:

Can anyone get Edge Copilot to work without logging in after July 2026?
https://copilot.microsoft.com/

As of this month, whenever I use MS Edge to go to Copilot, no matter what >>> tricks I use, it wants me to immediately log into a mothership account.

I can only find one trick that still works, and it only works in Edge.
But with no other browser can I get copilot to not ask for a login now.

Q: Other than sidebar Copilot, did you Copilot in Edge just recently ask >>> for a login to a mothership account every time you go to it using Edge?

NB: A separate thread on the Firefox ng shows nobody can do it there.
Newsgroups: alt.comp.software.firefox
Subject: Recently, did your Firefox stop using Microsoft Copilot in anonymous mode?
Date: Sun, 26 Jul 2026 14:00:07 -0700
Message-ID: <1145sgm$1cju$1@nnrp.usenet.blueworldhosting.com>

In Linux, I have a bookmark that I launch.
Now the first time I ran it tonight I had to log in, but the next 4 times I loaded Firefox
and launched Copilot, I didn't have to log in. Time would only tell.

Could it be that your Firefox has cookies set that tell it that you
are *still* logged in from the first session? "Keep me logged in" is
an option I sometimes see on sites.

It might be fun to try to log out of Copilot the clear the cookies
then restart a Firefox session and see if it asks for a login.

J.

I don't normally clear cookies, but I do have a script that selectively clears the storage
folder in Firefox. In the profile folder, storage->default->https+++copilot.microsoft.com
is the folder I exclude from deleting. Also in that folder is https+++account.microsoft.com. Again another I don't clear.

So this probably keeps some history?

--
Mint 22.3, Thunderbird 153.0esr, Firefox 153.0
Alan K.

--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Can anyone get Edge Copilot to work without logging in after July 2026?

On Tue, 7/28/2026 8:16 AM, Alan K. wrote:

On 7/27/26 11:14 PM, John wrote:

It might be fun to try to log out of Copilot the clear the cookies
then restart a Firefox session and see if it asks for a login.
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ ÿ J.

I don't normally clear cookies, but I do have a script that selectively clears the storage folder in Firefox.ÿ In the profile folder, storage->default->https+++copilot.microsoft.com is the folder I exclude from deleting.ÿ Also in that folder is https+++account.microsoft.com.ÿÿ Again another I don't clear.

So this probably keeps some history?

They don't really use cookies. You'll notice when a web page
accuses you of shutting off the cookies, that the cookies are
actually wide open.

What they are referring to, is the DOM storage, and the
abuse DOM storage receives is what they are after. Some older
browsers may not have working DOM storage or the Javascript
support to help abuse DOM storage.

Browser designs have been modified, to try to increase the
number of databases, and "cubby holes" needing cleaning.
That's the Chrome/Chromium imperative, is lots and lots
of databases that users will not be able to figure out.
Booby trap a few, so a user deletes them and the browser
stops working, and the job is complete.

*******

DOM storage uses three plus signs in the folder name "+++".
You can do a non-regex search for those, if you want
to delete them. There is a Permanent and a Temporary
DOM storage area, and some sites like a Youtube, may be
using both of those.

They have also used the webarchive database for some things,
but recent Google changes, they're not using that now.
Once that file gets up to 15MB or so, it may be slowing
down how web pages worked in the past.

Paul

--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Can anyone get Edge Copilot to work without logging in after July 2026?

Maria Sophia <mariasophia@comprehension.com> wrote:

Can anyone get Edge Copilot to work without logging in after July 2026?
https://copilot.microsoft.com/

Just use a local model and don't bother with any of the online ones. They
will *always* use you as training data unless you pay for a firewalled
model. Doesn't matter whether you login or not.

--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Can anyone get Edge Copilot to work without logging in after July 2026?

On 28/07/2026 17:53, Maria Sophia wrote:

Yup. The only time I ever set up an MSA was on a Windows S laptop, and that was before I learned you didn't need to create the MSA in order to convert
it to Windows 10, and even so, I deleted the account immediately (although Microsoft took a month to actually wipe it off the face of the earth).

Why did you do that? Was that account bothering you in any way?

You should always use a separate account to log in to CoPilot, ChatGPT,
and any other AI portals. While you don't need to use the email service,
you should log in occasionally to delete any unwanted emails. You don't
need to open any emails to do this. It's just a disposable account, so
it's highly unlikely that Microsoft or Google will ever send you $1,000
for using their services.


--
*Newsgroup Post *



--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Can anyone get Edge Copilot to work without logging in after July 2026?

On Tue, 7/28/2026 1:09 PM, Chris wrote:

Maria Sophia <mariasophia@comprehension.com> wrote:

Can anyone get Edge Copilot to work without logging in after July 2026?
https://copilot.microsoft.com/

Just use a local model and don't bother with any of the online ones. They will *always* use you as training data unless you pay for a firewalled
model. Doesn't matter whether you login or not.

The one I'm using, runs fine with the network cable pulled.
I think I've asked it three questions, since installing it
six months ago. Definitely the wave of the future. The
last question, took about four minutes to answer. This is a model
that according to the description, is "sharp at science, not so
good at current affairs".

One of the limits on the local model (something I haven't figured out yet),
is there is a claim of some sort of 4K token limit on input or output,
whereas the "model designs" claim to have 128K tokens as a limit. This means
in theory, I cannot drag and drop "War and Peace" into the query window.
The thing will still engage "chunking" at a guess, when a text is offered,
but the best way to offer text is via a web page URL. You could put some
text into a pastebin for example. And hope the machine knows where
the actual text sample is. Setting up a web server at 127.0.0.1
should work for that.

When a lot of things about running these are not explained,
there are bound to be some troubling observations. And you cannot
ask the machine anything that involves introspection. It doesn't
know what is in the training set, or, what it ate for lunch :-)
It's the very soul of disembodiment.

"Dave, my mind is going. I can feel it." [Not really]

Paul


--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Can anyone get Edge Copilot to work without logging in after July 2026?

On Tue, 7/28/2026 1:46 PM, Maria Sophia wrote:

That means my machine would likely only work with LM Studio or Ollama.
If I were to choose a setup for my old, 2009 Windows 10 box, it may be
a. LM Studio GUI
b. Phi 4 mini (3.8B) models
c. Which should work with ~2-3 GB disk & 16 GB RAM

For now, I don't have the time to research how to install a local LLM.
But if others are ahead of us, it would be good to know what they use.

It's simple. Install LMStudio.

Then, look for a model in the list of open weight models it has.
Go and research the options it offers, for the one you think
will work in your setup. You could start with a Gemma 2.4B or 4B
for example, which should run on just about anything.

Inference can be done purely on the CPU, and it doesn't even
necessarily make good usage of the CPU. I'm running mine on
the CPU, as my GPU is not good enough for most all of them.

The model description, if it is a mixture of experts,
will tell you how big one of the mixture elements is.
The download files could be 58GB, while the MOE might
take 20GB of VRAM. And that would tell you what kind
of video card could be used. For example, the biggest Kimi
model, is 100GB+ for VRAM, so an RTX6000 at 96GB would not
be enough. It would take two RTX6000, and you have to
go off and research whether your video card choice supports
"sharding". Sharding does not apparently always work well.
And depending on the video card brand, may not work at all.
If a mixture of experts had different sizes models, it might
manage to run one model per video card.

It takes maybe a minute to load a large model, and it is
not purely I/O constrained. Initially it works at
the I/O rate, but once it gets to about 80% of the model
loaded, things slow down for some reason.

You don't need to do much, to get a text query window running.
It will be a lot harder, to research what exactly it is doing.
At the lower level, there is matrix math. At a higher level,
there is a strategy planner, and an ordered list of Experts
to run. And a few models, do not use a strategy planner, and
they bind an Expert to each token. The output rate on that one,
is well below 1 token per second, the approach then being a
"novelty" rather than being practical.

Paul

--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Can anyone get Edge Copilot to work without logging in after July 2026?

Maria Sophia <mariasophia@comprehension.com> wrote:

Chris wrote:

Maria Sophia <mariasophia@comprehension.com> wrote:

Can anyone get Edge Copilot to work without logging in after July 2026?
https://copilot.microsoft.com/

Just use a local model and don't bother with any of the online ones. They
will *always* use you as training data unless you pay for a firewalled
model. Doesn't matter whether you login or not.

Thanks for that idea, which, I agree, is the best way to run an LLM.

I researched the local models and they're perfect except... except that
they take a ton of disk space and seem to require a beefy machine.

Dunno if my 2009 Windows 10 desktop can handle a local model but I do agree that's the way to go for someone like me who cares about privacy.

I'll have to look into it 'cuz these web-page LLMs have gotten me hooked on the convenience, but, one by one, they've each started requiring a login.

Well, then. Looks like you're going to have put a price on your desire for "privacy".

Looking it up, while other families exist (e.g., Qwen, DeepSeek, Gemma, Mistral), apparently Llama remains the most widely used on Windows.

Disk space needed seems to be about 3-7 GB for an 8 billion parameters LLM model, and maybe around 30-40 GB for a 70 billion parameters LLM model.

Parameters are the learned weights inside the neural network, where more parameters add deeper reasoning and stronger language ability.

Apparently disk space depends on quantization (i.e., compression) though.

Looking it up...
The big problem is RAM where 64­V128 GB RAM is normal for local LLMs.
The GPU needs to have VRAM starting at around 48 GB for local LLMs too.

Apparently nost consumer GPUs (RTX 3060/3070/3080/4070/4080) will not work
as we need workstation-class cards (i.e., RTX 6000 Ada, A100, H100, etc.).

Lol no. That's the spec for *training* an LLM. Running an LLM requires a
lot less.


--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Can anyone get Edge Copilot to work without logging in after July 2026?

Same without using Edge. MS really want users to log in. :(


In alt.comp.microsoft.windows Maria Sophia <mariasophia@comprehension.com> wrote:

Can anyone get Edge Copilot to work without logging in after July 2026?
https://copilot.microsoft.com/

As of this month, whenever I use MS Edge to go to Copilot, no matter what tricks I use, it wants me to immediately log into a mothership account.

I can only find one trick that still works, and it only works in Edge.
But with no other browser can I get copilot to not ask for a login now.

Q: Other than sidebar Copilot, did you Copilot in Edge just recently ask
for a login to a mothership account every time you go to it using Edge?

NB: A separate thread on the Firefox ng shows nobody can do it there.
Newsgroups: alt.comp.software.firefox
Subject: Recently, did your Firefox stop using Microsoft Copilot in anonymous mode?
Date: Sun, 26 Jul 2026 14:00:07 -0700
Message-ID: <1145sgm$1cju$1@nnrp.usenet.blueworldhosting.com>

--
"I press on toward the goal to win the prize for which God has called me heavenward in Christ Jesus." --Philippians 3:14. When does this old ant get it? Note: A fixed width font (Courier, Monospace, etc.) is required to see this signature correctly.
Note: A fixed width font (Courier, Monospace, etc.) is required to see this signature correctly.
/\___/\ Ant(Dude) @ http://aqfl.net & http://antfarm.home.dhs.org.
/ /\ /\ \ Please nuke ANT if replying by e-mail.
| |o o| |
\ _ /
( )

--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Can anyone get Edge Copilot to work without logging in after July 2026?

On Tue, 7/28/2026 4:43 PM, Chris wrote:

Maria Sophia <mariasophia@comprehension.com> wrote:

Chris wrote:

Maria Sophia <mariasophia@comprehension.com> wrote:

Can anyone get Edge Copilot to work without logging in after July 2026? >>>> https://copilot.microsoft.com/

Just use a local model and don't bother with any of the online ones. They >>> will *always* use you as training data unless you pay for a firewalled
model. Doesn't matter whether you login or not.

Thanks for that idea, which, I agree, is the best way to run an LLM.

I researched the local models and they're perfect except... except that
they take a ton of disk space and seem to require a beefy machine.

Dunno if my 2009 Windows 10 desktop can handle a local model but I do agree >> that's the way to go for someone like me who cares about privacy.

I'll have to look into it 'cuz these web-page LLMs have gotten me hooked on >> the convenience, but, one by one, they've each started requiring a login.

Well, then. Looks like you're going to have put a price on your desire for "privacy".

Looking it up, while other families exist (e.g., Qwen, DeepSeek, Gemma,
Mistral), apparently Llama remains the most widely used on Windows.

Disk space needed seems to be about 3-7 GB for an 8 billion parameters LLM >> model, and maybe around 30-40 GB for a 70 billion parameters LLM model.

Parameters are the learned weights inside the neural network, where more
parameters add deeper reasoning and stronger language ability.

Apparently disk space depends on quantization (i.e., compression) though.

Looking it up...
The big problem is RAM where 64­V128 GB RAM is normal for local LLMs.
The GPU needs to have VRAM starting at around 48 GB for local LLMs too.

Apparently nost consumer GPUs (RTX 3060/3070/3080/4070/4080) will not work >> as we need workstation-class cards (i.e., RTX 6000 Ada, A100, H100, etc.).

Lol no. That's the spec for *training* an LLM. Running an LLM requires a
lot less.

A Google, seems to dig up tables of such info.

https://www.spheron.network/blog/gpu-memory-requirements-llm/

Quick Answer: How Much VRAM Do You Need for an LLM?

As a rule of thumb, an LLM needs about 2 GB of VRAM per billion parameters at FP16,
or roughly 0.5 GB per billion at INT4, plus 15-20% on top for the KV cache, activations,
and framework overhead.

So a 7B model fits in ~16-20 GB, a 70B model needs ~140-170 GB at FP16 (or ~46 GB at INT4),
and a 685B model like DeepSeek V3.2 needs ~822 GB at FP8.

Model size FP16 VRAM INT4 VRAM Smallest GPU setup that runs it
7-8B ~16-20 GB ~6-7 GB 1x RTX 4090 (24 GB)
13-14B ~34 GB ~10 GB 1x RTX 5090 (32 GB), INT4
32B ~76 GB ~22 GB 1x RTX 4090/5090 (24-32 GB), INT4

70-72B ~168-172 GB ~46-47 GB 1x L40S (48 GB) at INT4, or 1x B300 at FP16
109-123B ~262-295 GB ~66-75 GB 1x H100 (80 GB), INT4

400B+ MoE ~960 GB+ ~240 GB+ 1x B300 (288 GB) or 4x H100, INT4 {forgetaboutit territory]
685B MoE ~1.6 TB ~411 GB 8x H100 / 8x H200 at FP8

The same web site, has this on a separate page.

GPU Requirements: VRAM, Memory, and Storage
Variant Precision VRAM
GPT-OSS 20B BF16 ~42 GB
GPT-OSS 20B FP8 ~21 GB
GPT-OSS 120B MoE MXFP4 ~60 GB <=== running on Big Machine 128GB system RAM (no GPU) [58GB download]
GPT-OSS 120B MoE BF16 TP4 ~240 GB

*******

My recommendation, is start with a small model, like a Gemma 2.4B or 4B, see how much RAM it takes to run, as a "calibration" of some of these tables.

On Linux, the LMStudio was an AppImage. Just download it and run it.
AppImage is a kind of container, similar to FlatPak or SNAP.

This is not the latest, I checked the browser on the machine, to see where
I got the AppImage from. This would be a few months out of date.

https://installers.lmstudio.ai/linux/x64/0.4.12-1/LM-Studio-0.4.12-1-x64.AppImage

The Harmony package isn't theirs, so it comes from somewhere as well. You will see that item listed in my downloads, in the picture below.

In any case, you don't have to know a lot to use it. That's the whole idea. Load an open-weights model, give it a try. In this case, with the 60GB model loaded, system RAM consumption is in the 80GB range. So there's a bit of overhead on system RAM. If the Mixture of Experts were loaded to a video card, the size for that is less, but it's generally going to bust any video card
you can afford. This is where the machines with a low end iGPU that maps
96GB of system RAM for the GPU, comes in handy, as you buy a 128GB machine for $3K-$6K, it isn't as fast as an RTX6000, but it also doesn't cost $15K just
for the video card with 96GB of fast VRAM. Apple this week, is apparently playing with the idea of leasing things like a Mac Studio, as a solution
for the RAM cost/availability issue.

[Picture] Paul uses his LLM-AI...

https://i.postimg.cc/qqXCTQ7V/LMStudio-memory-usage.gif

https://imgur.com/a/1s2JZcv

A short sampling of the C code it wrote. Yes, it cheated.
What was I thinking :-) Oh, yeah, this was a simulation.

/* -------------------------------------------------------------
* 2) Convert decimal digits ? binary precision.
*
* bits ÷ digits ? log?(10)
* Add a safety margin (64?bits ÷ 20 extra decimal digits).
* ------------------------------------------------------------- */
const double LOG2_10 = 3.32192809488736234787; /* exact enough */
mpfr_prec_t prec = (mpfr_prec_t)ceil(dec_digits * LOG2_10) + 64;

/* -------------------------------------------------------------
* 2) Initialise MPFR with the required precision.
* ------------------------------------------------------------- */
mpfr_set_default_prec(prec);
mpfr_t pi;
mpfr_init(pi); /* allocates space for a variable of 'prec' bits */

/* -------------------------------------------------------------
* 3) Compute ã. MPFR's const_pi uses a very fast AGM?based method <=== as a student, it always cheated at school...
* and returns a correctly rounded value (round?to?nearest). My fault for not slapping it around a bit...
* ------------------------------------------------------------- */
mpfr_const_pi(pi, MPFR_RNDN); /* MPFR_RNDN = round to nearest */

With a good deal more effort, I could convince it to use
one of the better PI algo that have been invented. Instead
of using an infinite precision library that happens to have available
to it, the "constant" PI. Cheating with ArcTan, might take longer.

Paul

--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Subject: Re: Can anyone get Edge Copilot to work without logging in after July 2026?

On Tue, 7/28/2026 8:07 PM, Ant wrote:

Same without using Edge. MS really want users to log in. :(

In alt.comp.microsoft.windows Maria Sophia <mariasophia@comprehension.com> wrote:

Can anyone get Edge Copilot to work without logging in after July 2026?
https://copilot.microsoft.com/

As of this month, whenever I use MS Edge to go to Copilot, no matter what
tricks I use, it wants me to immediately log into a mothership account.

I can only find one trick that still works, and it only works in Edge.
But with no other browser can I get copilot to not ask for a login now.

Q: Other than sidebar Copilot, did you Copilot in Edge just recently ask
for a login to a mothership account every time you go to it using Edge?

NB: A separate thread on the Firefox ng shows nobody can do it there.
Newsgroups: alt.comp.software.firefox
Subject: Recently, did your Firefox stop using Microsoft Copilot in anonymous mode?
Date: Sun, 26 Jul 2026 14:00:07 -0700
Message-ID: <1145sgm$1cju$1@nnrp.usenet.blueworldhosting.com>

That's understandable, once you consider the attack surface
these projects offer to the not-so-honest public.

1) Commercial usage of a consumer test item.
They also offer a commercial version, paid tokens, and they have
to implement a usage counter to stop abuse. If you do enough free
queries, I bet you'll get stopped eventually.

2) It's a lot like running a USENET News server without authentication,
and then being surprised when the server is filled with sporge or a flood.
Authentication makes it easier to track abuse or note abuse patterns.
Like when Kimi is scraping your datacenter. Or OpenAI is attacking Huggingface.
Or... you get the idea.

I don't particularly blame them for having some amount of tracking
on the activity.

You can always run a local model, if you don't like this. Scroll down, to the text section.

https://lmstudio.ai/docs/app/system-requirements

Start with a small model, and see if the speed bothers you. I get 9.7 tokens per second
or so, as an output speed. The best machine you can buy, is estimated
to produce 3000 tokens per second, or 300x faster roughly. That's what
running locally (without a GPU...) buys you. Not all the machines Microsoft runs, are even remotely that fast. They have experimented with inference
on "old beaters", less than optimal boxes. I've had a couple sessions
that seemed excessively slow. But that's what I expect them to do, is
calibrate the service. Sometimes the thinking stage is so fast, you
don't even know half the output tokens have already been emitted,
and you have to scroll down to see how much progress it has made.

And I don't do a lot of these local sessions. For this discussion thread,
I have done my fourth query, which was to write a C program, which it did.
I did not constrain the problem enough, for a practical output. The
problem is, I attempted to "type in a one-liner", and that's a mistake.

And this is a reminder about working style.

Take a text editor. Prepare your query. Add guard statements and
anti-cheat text. Think about all the holes that are left in the
query. Spend ten minutes if you have to...

Then, paste the paragraph of text, into the query box. Press
the "Up Arrow" icon on the right of the pasted text, to start the query.

Preparing the query in an offline manner, helps you write better
queries. Which is important if you are paying for the tokens.
"If you shoot from the hip, you get shit for answers..."

Paul


--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)