---
title: 'Zed Editor + Local LLMs: Full Setup Guide with Ollama and LM Studio'
source: 'https://youtube.com/watch?v=Bq2po2T3h7k'
video_id: 'Bq2po2T3h7k'
date: 2026-06-14
duration_sec: 386
---

# Zed Editor + Local LLMs: Full Setup Guide with Ollama and LM Studio

> Source: [Zed Editor + Local LLMs: Full Setup Guide with Ollama and LM Studio](https://youtube.com/watch?v=Bq2po2T3h7k)

## Summary

This video demonstrates how to connect the Zed editor to local LLMs using Ollama and LM Studio, both on the same machine and across a local network. The process is straightforward and requires only a few configuration steps.

### Key Points

- **Verifying Ollama Setup** [0:35] — Check that Ollama is working by listing models and running a test query in the console.
- **Configuring Ollama in Zed** [1:03] — In Zed's agent panel settings, select Ollama provider, adjust context window size (e.g., 60,000 tokens), and click connect.
- **Agent Profile and Thinking Mode** [2:22] — Choose the appropriate agent profile for file editing capabilities. Toggle thinking mode for reasoning models to speed up responses.
- **Setting Up LM Studio** [2:55] — Similar to Ollama, verify LM Studio works in console, then in Zed select LM Studio provider and connect.
- **Connecting to Remote LLMs** [4:15] — Add a new OpenAI-compatible provider with the remote machine's URL and port. Ensure the full path including protocol version is specified.

### Conclusion

Connecting local LLMs to Zed editor is simple and flexible, allowing seamless switching between models on the same machine or across a local network.

## Transcript

One of the editors I use is Zed editor,
and most of the time I use it together
with local models. So today I want to
show you how I connect it to local
LLM's, both models running on the same
computer as the editor, and models
running on another computer in the local
network that can be very convenient when
you want to run models on a more
powerful machine with a stronger GPU.
And since I usually use Ollama and LM
Studio to run models, I'll use those as
examples today. But before that, I want
to congratulate the Zed editor team on
the release of version 1.0. I've been
using this editor since the closed beta
days, and the progress over the last few
years has been truly impressive. So
let's start with Ollama. First, I'll
check in the console that Ollama is
working properly, so if anything goes
wrong later, I know the issue is not
there. First, let's look at the list of
models available on this computer. I
already have several downloaded. Let me
choose Gemma 4. Now I'll launch it and
ask it to answer something, just to make
sure it works correctly. Yes, the model
loaded into memory successfully, thought
for a moment, and then gave me a
response. So everything looks good. Now
I know that both Ollama and the selected
model are working. Now let's go to Zed
editor and open the agent panel. In the
settings for this panel, there is a
section for LLM providers. That's
exactly what we need. There we can find
an option called Ollama. By default, the
local URL with the correct port is
already filled in here. That is the
default Ollama address, and if you
didn't change anything when setting up
Ollama, then you don't need to change
anything here either. I will only
increase the context window size because
when the model uses tools, 8,000 tokens
may be too little. I'll increase it to
60,000 for now. That should be enough to
start with. Now we press the connect
button so Zed editor can verify that
Ollama is available at that URL and load
the list of available models. As you can
see, a green check mark appeared along
with the word connected, which means Zed
successfully connected to Ollama, and
everything is working properly. All
right, now let's close the settings
window and look at the list of available
models. We can see a new section called
Ollama, and inside it are the models
available to us. As you can see, they
are the same models that were shown in
the console earlier when we launched
Ollama. Let me also choose the same
model here that I used in the console,
Gemma 4. And again, just to verify that
everything is definitely working, I'll
type something to it. It doesn't really
matter what. The important thing is to
confirm that I can interact with my
local model directly from the editor.
Okay, the model thought for a moment and
then replied. Great, that means
everything is configured and working
exactly as it should. Let me also show
you a couple of important things you
should pay attention to. The first one
is the agent profile. To simplify it a
lot, this determines what the agent is
allowed to do. There are three built-in
profiles, and if you want the model to
be able to edit and create files inside
your project, choose the right profile.
The second thing I want to point out is
the thinking mode switch. If the LLM you
selected supports reasoning before
answering, you can enable or disable it
here. It can be useful to experiment
with turning it on or off, especially if
the model responds very slowly and you
want to speed things up a bit. All
right, now that we've covered Ollama,
let's do the same thing with LM Studio.
Just like with Ollama, I'll first check
in the console that LM Studio starts
without any problems. I'll also launch a
model to make sure everything is working
properly there as well. Since I
currently have only one model loaded in
LM Studio, that's the one I'll use. Yes,
everything works here too. The model
responds to my requests. Okay, now let's
go back to Zed editor and once again
open the agent panel settings section
with LLM providers. This time we look
for the option called LM Studio. Here
again, just like with Ollama, the
correct default URL is already entered,
so there is nothing to change. We simply
click connect, and Zed editor checks
that everything is working, connects to
LM Studio, and retrieves the list of
available models. As you can see,
everything went successfully. The editor
connected to LM Studio. Now let me open
a new chat and then choose a new model
for that chat provided by LM Studio. As
you can see, a new section dedicated to
LM Studio has appeared, and since I only
have one model loaded in LM Studio,
there is not much to choose from. So we
simply select that one. Okay, this time
I'll ask the model to do something
useful, and as you can see, it
successfully processed my request and
wrote "Hello world" in several
languages. Excellent. The combination of
Zed editor and LM Studio works as well.
So, as you can see, connecting local
models to Zed editor is quite simple. It
only takes a few clicks, and by the way,
you can switch between them at any time.
Now let me show you a slightly more
advanced case, when the LLM's are
running on another computer inside your
local network. To do that, we go into
the model settings. The simplest way
would be to just change the URL for the
existing Ollama and LM Studio entries.
If you only plan to use models running
on another computer, then that is
exactly what you should do. But I want
to be able to switch dynamically between
smaller models running on my laptop and
larger models running on my desktop
machine. So I'll use a different
approach. In the header of the LLM
providers section, there is a button for
adding a new provider. We click it and
choose the option OpenAI compatible.
Then we enter a name. In my case, I'll
call it Ollama desktop, so I always know
what these models are. And then I enter
the address of the other computer where
Ollama is running.
In the API key field, you can enter any
text. By default, Ollama does not
require a key. After that, we save it.
Oh, wait, it won't let me save without
specifying the default model name. All
right, let me enter one of the models I
already have loaded on my desktop. Okay,
now that everything is filled in and
saved, we can finally test it. I'll type
a test message in the chat.
Oops, it doesn't work. Looks like I made
a mistake somewhere. But that's fine,
we'll figure it out. Let's close the
agent panel for a moment and open the
settings file so it's easier to see
where I went wrong. By the way, this is
a very convenient feature in Zed editor.
Every change you make through the UI is
reflected directly in the config files.
All right, here is the section for the
provider I just added. So what's wrong
here? Ah, I see it. In the API URL
field, when adding your own LLM
provider, you need to specify the full
path including the port number and the
protocol version. All right, I'll save
the settings and try again. Success. Now
it works. I was able to access an LLM
running on my desktop from the editor
running on my laptop. Excellent. I'm not
going to demonstrate connecting to LM
Studio on the desktop because the
process is identical to Ollama, only the
URL is slightly different. So as you can
see, using local models in Zed editor is
very simple. And on top of that, you can
configure everything in whatever way
works best for you. That's exactly why I
like Zed so much. That's all for today.
If you enjoyed this video, give it a
like and subscribe so you don't miss the
next episodes. See you soon. Take care.