It seems we might need more instructions on how to set up python (via uv) in vanilla MacOS.
ontouchstart 2 days ago [-]
I have been updating the gist above and will stop this
```
../scipy/meson.build:274:9: ERROR: Dependency lookup for OpenBLAS with method 'pkgconfig'
failed: Pkg-config for machine host machine not found. Giving up.
```
Too much work.
zokier 3 days ago [-]
What is unsloths business/income? They seem to be publishing lot of stuff for free, with no clear product to back them?
danielhanchen 2 days ago [-]
Hey! Our primary objective for now is to provide the open source community with cool and useful tooling - we found closed source to be much more popular because of better tooling!
We have much much in the pipeline!!
brainless 2 days ago [-]
Thanks! How do you earn or keep yourself afloat? I really like what you guys are doing. And similar orgs. I am personally doing the same, full-time. But I am worried when I will run out of personal savings.
vessenes 19 hours ago [-]
I've been wondering this since they started it, mostly as a concern they stay afloat. Since Daniel does the work of ten, it seems like their value:cost ratio is world-class at the very least.
With the studio release, it seems to like they could be on the path to just bootstrapping a unicorn or a 10x corn or whatever that's called, which is super interesting. Anyway, his refusal to go into details reassures me, sounds like things are fine, and they're shipping. Vai com dios
richardw 2 days ago [-]
Daniel is a very impressive guy. Well within the realm of “fund the people not the idea” that YC seems to do. Got a few bucks from them and probably earning from collaborations etc. Odds of them not figuring out a business model seem slim.
From comments elsewhere in this thread, it sounds like Unsloth could also be getting some decent consulting revenue from larger companies.
reactordev 2 days ago [-]
The opportunity here is HUUUUGGGEEEE!!!
Companies have no idea what they are doing, they know they need it, they know they want it, engineers want it, they don’t have it in their ecosystem so this is a perfect opportunity to come in with a professional services play. We got you on inference training/running, your models, all that, just focus on your business. Pair that with huggingface’s storage and it’s a win/win.
zokier 2 days ago [-]
Investments are not income
csomar 2 days ago [-]
You didn't answer the parent question.
segmondy 2 days ago [-]
They don't owe anyone an answer.
zokier 2 days ago [-]
But if they want to attract users, like they seem to do, then answering would go long way.
kartaka83838 2 days ago [-]
that doesnt sound reassuring?
2 days ago [-]
throwa356262 2 days ago [-]
With a team size of eight (!) I think they are not exactly bleeding money
jacquesm 2 days ago [-]
Unsloth is the real thing. Highly recommended for those running their own AI engines and that want to get the most out of them.
illusive4080 2 days ago [-]
Apache license. Can’t wait to try it out at work! LMStudio’s proprietary license makes getting permission hard.
halJordan 2 days ago [-]
Some of it is Apache
smcnally 2 days ago [-]
Some of unsloth studio’s code is Apache? Or some of lmstudio is?
mkl 2 days ago [-]
From the README at https://github.com/unslothai/unsloth: "Unsloth uses a dual-licensing model of Apache 2.0 and AGPL-3.0. The core Unsloth package remains licensed under Apache 2.0, while certain optional components, such as the Unsloth Studio UI are licensed under AGPL-3.0."
illusive4080 2 days ago [-]
Good to know. AGPL is fine at my workplace too, whereas the custom LMStudio license has been in review since mid January with no ETA in sight.
solomatov 2 days ago [-]
What do you mean by custom LMStudio license? Your employer requires reviews of proprietary EULAs or do you try to get a custom licensing deal from LMStudio?
illusive4080 1 days ago [-]
Employer must review all EULA terms, yes. The license is a hand crafted proprietary license not a standard OS license.
keyle 2 days ago [-]
I noticed the training part, that was interesting for my use case. Unfortunately it's said 'on NVIDIA'.
Is there an alternative, tutorial, or project you'd recommend that would help me do supervised fine tuning (SFT) with the metal stack / macOS?
MrDrMcCoy 2 days ago [-]
They state further down that they're working on non-Nvidia support. Looking forward to it, since I'm pretty heavily invested in suffering on AMD (ROCm sucks, but everything else about AMD is worth it to me.).
syntaxing 2 days ago [-]
> Mac: Like CPU - Chat only works for now. MLX training coming very soon
TiredOfLife 2 days ago [-]
Use that mac to rent Nvidia gpus
kahrl 2 days ago [-]
lol.
car 3 days ago [-]
Tried to build from source on MacOS, but got this error:
(base) unsloth git:(main) unsloth studio setup
╔══════════════════════════════════════╗
║ Unsloth Studio Setup Script ║
╚══════════════════════════════════════╝
Node v25.8.1 and npm 11.11.0 already meet requirements. Skipping nvm install.
Node v25.8.1 | npm 11.11.0
npm run build failed (exit code 2):
> unsloth-theme@0.0.0 build
> tsc -b && vite build
src/features/chat/shared-composer.tsx(366,17): error TS6133: 'status' is declared but its value is never read.
danielhanchen 2 days ago [-]
Hey will check ASAP and fix - sorry about that
danielhanchen 2 days ago [-]
Apologies on the delay - we fixed it! Please re-try with:
Thank you for the follow up! Big fan of your models here, thanks for everything you are doing!
Works fine on MacOS now (chat only).
On Ubuntu 24.04 with two GPU's (3090+3070), it appears that Llama.cpp sometimes uses the CPU and not GPU. This is judging from the tk/s and CPU load for identical models run with US-studio vs. just Llama.cpp (bleeding edge).
pentagrama 1 days ago [-]
LM Studio user here. Unsloth looks great, wanted to check it out, but there is no app file to download and install? Sorry I'm not familiar with the command line.
trvz 3 days ago [-]
Installing with pip on macOS is just not an acceptable option. It'll mess up your system just like npm or gem.
This needs to go on homebrew or be a zip file with an app for manual download.
DIVx0 3 days ago [-]
Agree with you, a slightly more maintainable way to use it now is with "uv" or mise. i've used `uv tool install unsloth` for this one.
danielhanchen 3 days ago [-]
Yep - uv is a better fit - and you get parallel downloads as well
saagarjha 2 days ago [-]
It's not just uv, it also will drop an nvm install in your home directory :(
danielhanchen 3 days ago [-]
Hey we're still working on making installation much better - appreciate the feedback!
We come from Python land mainly so packaging and distribution is all very new to us - homebrew will definitely be next!
mmis1000 2 days ago [-]
Uv helps you up though. Use a pyproject.toml and uv sync. Everything will be put into the venv only, nothing spread across the whole system.
The pyproject.toml can even handles build env for you, so you no longer need a setup.sh that installs 10 tool in specific order with specific flag to produce working environment. A single uv sync, and the job is done.
Plus the result is reproducible, so if this time uv sync work, then it also work next time.
Highly recommend if you are still on pip.
Note: Take a example that I used to install unsloth with rocm setup that based on unreleased git version dependencies and graphic card specific build flag, all of them can be handled with one command 'uv sync'. This will require a big pile of shell script if doing another way. https://github.com/unslothai/unsloth/issues/4280#issuecommen...
bityard 3 days ago [-]
I recommend installing uv first, then you can install any Python code you want inside a virtual environment to keep it isolated from the rest of the system.
danielhanchen 3 days ago [-]
Yep uv pip install unsloth works as well - we probably should have just made that the default - in fact Unsloth makes its own venv using UV if you have it dynamically
mmis1000 2 days ago [-]
I think the website should probably mention those installation preset in unsloth pyproject.toml though. The website instruct you to install dependencies separately. But it turns out there are dedicated preset that install specific rocm/cuda/xpu version in the project.
anentropic 2 days ago [-]
or `uv tool install unsloth` for a safe 'global' installation
Computer0 3 days ago [-]
On my linux systems I use venv to not affect system packages, is that not an option for this situation?
I know the whole package system across most languages is a dumpster fire but for Python, uv solves a lot of problems.
uv init
uv add unsloth
uv run main.py % or whatever
danielhanchen 3 days ago [-]
Yep UV is fantastic - should have just that default!
smcleod 2 days ago [-]
Ah yes, came to say something similar, Python dependencies are an absolute nightmare, even with uv it feels like there's always a battle to make other peoples Python apps install.
Update: It looks like it doesn't work with the current Python version, you might have to downgrade to Python 3.13 (however even then I still get `error: unexpected argument '--torch-backend' found`)
beernet 3 days ago [-]
Agreed, feels like a vibe-coded frontend based on already given backend features.
Also, never saw any Unsloth related software in production to this day. Feels strongly like a non-essential tool for hobby LLM wizards.
danielhanchen 3 days ago [-]
You would be surprised - we're the 4th largest independent distributor of LLMs in the world - and nearly every Fortune 500 company has utilized either our RL fine-tuning package or used our quants and models - we for example collab directly with large labs to release models with bug fixes.
airspresso 2 days ago [-]
Unsloth is providing the best and most reliable libraries for finetuning LLMs. We've used it for production use-cases where I work, definitely solid.
danielhanchen 2 days ago [-]
Glad it was helpful!
ta9000 2 days ago [-]
Even a brief reading of their site would have spared you this embarrassment.
kuon 2 days ago [-]
I hope you will support AMD. This looks Nice but I went the complicated route with AMD GPUs.
MrDrMcCoy 2 days ago [-]
They state that they're working on it. I'm waiting, too.
claaams 2 days ago [-]
Will check back when there's AMD support.
meowokIknewit 3 days ago [-]
The GUI for the fine tuning looks interesting. Hopefully this leads to a lot of new custom models
danielhanchen 3 days ago [-]
Thank you! We're still iterating on it so any suggestions are welcome!
yomansat 2 days ago [-]
FYI, if any devs are around, the privacy policy still links to the gitbook.
danielhanchen 2 days ago [-]
Oh will check and fix - thanks
saberience 3 days ago [-]
Who's the intended user for this?
Is it like, for AI hobbyists? I.e. I have a 4090 at home and want to fine-tune models?
Is it a competitor to LMStudio?
danielhanchen 2 days ago [-]
You would be surprised! Nearly every Fortune 500 company has utilized either our RL fine-tuning package or used our quants and models - the UI was primarily a culmination of pain points folks had when doing either training or inference!
We're complimentary to LM Studio - they have a great tool as well!
TheTaytay 2 days ago [-]
I don’t know why this is being downvoted. Danielhanchen is legit, and unsloth was early to the fine-tuning on a budget party.
danielhanchen 2 days ago [-]
Haha no worries at all :)
lastdong 3 days ago [-]
From the homepage looks like it:
“Training: Works on NVIDIA GPUs: RTX 30, 40, 50, Blackwell, DGX Spark/Station etc.”
mlnj 3 days ago [-]
I am unaware lm studio is being used for fine tuning. I believe it only does inference.
Happy to see unsloth making it even easier for people like me to get going with fine tuning. Not that I am unable to I'm just lazy.
Fine tuning with a UI is definitely targeted towards hobbyists. Sadly I'll have to wait for AMD ROCm support.
danielhanchen 2 days ago [-]
Thanks! We do have normal AMD support for Unsloth but yes the UI doesn't support it just yet! Will keep you posted!
MrDrMcCoy 2 days ago [-]
What does "normal AMD support" mean here? I was completely unable to get it working on my Ryzen AI 9700 XT. I had to munge the versions in the requirements to get libraries compatible with recent enough ROCm, and it didn't go well at all. My last attempt was a couple weeks before studio was announced.
huydotnet 3 days ago [-]
you just answered your own question, "AI hobbyists who has 4090 at home". And they are pretty much targeted user of Unsloth since the start.
danielhanchen 2 days ago [-]
Actually the opposite haha- more than 50% of our audience comes from large organizations eg Meta, NASA, the UN, Walmart, Spotify, AWS, Google, and the list goes on!
fl0id 2 days ago [-]
Does it use gpu support on mac? At least when running setup, it said cpu only.
2 days ago [-]
jawerty 3 days ago [-]
Excited to use this been using unsloth models for the past couple years
danielhanchen 3 days ago [-]
Thank you for your continued support - we have much more planned for it!
2 days ago [-]
Havoc 2 days ago [-]
Congrats to team. Busy build a dataset for training so will def test it out
peddling-brink 2 days ago [-]
This looks really cool. Any chance you'll support pretraining runs as well?
ma2kx 2 days ago [-]
Nice! Is there something planned to run the finetuning via hf jobs or runpod?
car 3 days ago [-]
Can Unsloth Studio use already downloaded models?
nodja 2 days ago [-]
IDK how it did but it detected my LM studio downloaded models I have on a spinning drive (they're not in the default location).
yacin 2 days ago [-]
wish there were an option to disable the annoying startup messages with emojis when using the library.
However, since I already have pi working with llama.cpp server from a docker container, I did a quick experiment to compare three code bases:
https://gist.github.com/ontouchstart/7483c12efa3c3d3a49e38c2...
https://gist.github.com/ontouchstart/217fe2b8103a5c0bfaee1e9...
Very interesting.
Will do it again next week if I can get unsloth studio working.
curl -LsSf https://astral.sh/uv/install.sh | sh
uv venv unsloth_studio --python 3.13
source unsloth_studio/bin/activate
uv pip install unsloth==2026.3.7 --torch-backend=auto
unsloth studio setup
unsloth studio -H 0.0.0.0 -p 8888
https://gist.github.com/ontouchstart/532312fcba59aec3ce7f6aa...
Here is the error message on my machine:
https://gist.github.com/ontouchstart/86ca3cbd8b6b61fa0aeec75...
It seems we might need more instructions on how to set up python (via uv) in vanilla MacOS.
``` ../scipy/meson.build:274:9: ERROR: Dependency lookup for OpenBLAS with method 'pkgconfig' failed: Pkg-config for machine host machine not found. Giving up. ```
Too much work.
We have much much in the pipeline!!
With the studio release, it seems to like they could be on the path to just bootstrapping a unicorn or a 10x corn or whatever that's called, which is super interesting. Anyway, his refusal to go into details reassures me, sounds like things are fine, and they're shipping. Vai com dios
https://www.ycombinator.com/companies/unsloth-ai
Companies have no idea what they are doing, they know they need it, they know they want it, engineers want it, they don’t have it in their ecosystem so this is a perfect opportunity to come in with a professional services play. We got you on inference training/running, your models, all that, just focus on your business. Pair that with huggingface’s storage and it’s a win/win.
Is there an alternative, tutorial, or project you'd recommend that would help me do supervised fine tuning (SFT) with the metal stack / macOS?
curl -LsSf https://astral.sh/uv/install.sh | sh
uv venv unsloth_studio --python 3.13
source unsloth_studio/bin/activate
uv pip install unsloth --torch-backend=auto
unsloth studio setup
unsloth studio -H 0.0.0.0 -p 8888
Works fine on MacOS now (chat only).
On Ubuntu 24.04 with two GPU's (3090+3070), it appears that Llama.cpp sometimes uses the CPU and not GPU. This is judging from the tk/s and CPU load for identical models run with US-studio vs. just Llama.cpp (bleeding edge).
This needs to go on homebrew or be a zip file with an app for manual download.
We come from Python land mainly so packaging and distribution is all very new to us - homebrew will definitely be next!
The pyproject.toml can even handles build env for you, so you no longer need a setup.sh that installs 10 tool in specific order with specific flag to produce working environment. A single uv sync, and the job is done.
Plus the result is reproducible, so if this time uv sync work, then it also work next time.
Highly recommend if you are still on pip.
Note: Take a example that I used to install unsloth with rocm setup that based on unreleased git version dependencies and graphic card specific build flag, all of them can be handled with one command 'uv sync'. This will require a big pile of shell script if doing another way. https://github.com/unslothai/unsloth/issues/4280#issuecommen...
https://pipx.pypa.io/stable/installation/
uv init
uv add unsloth
uv run main.py % or whatever
Update: It looks like it doesn't work with the current Python version, you might have to downgrade to Python 3.13 (however even then I still get `error: unexpected argument '--torch-backend' found`)
Also, never saw any Unsloth related software in production to this day. Feels strongly like a non-essential tool for hobby LLM wizards.
Is it like, for AI hobbyists? I.e. I have a 4090 at home and want to fine-tune models?
Is it a competitor to LMStudio?
We're complimentary to LM Studio - they have a great tool as well!
Happy to see unsloth making it even easier for people like me to get going with fine tuning. Not that I am unable to I'm just lazy.
Fine tuning with a UI is definitely targeted towards hobbyists. Sadly I'll have to wait for AMD ROCm support.