mongrelion

letting the market set prices ensures that the chips go to the critical markets and uses.

Can you please elaborate what you mean by "critical market"?

Edit: formatting

llama.cpp moves too quickly to be added as a stable package. Instead, you can get it directly from AUR: https://aur.archlinux.org/packages?O=0&K=llama.cpp

There are packages for Vulkan, ROCm and CUDA. They all work.

I have been so far happy with the value that Copilot brought but for the past few weeks I have felt the chokehold on the number of requests.

I have had the chance to test the main Chinese models through OpenRouter but the Pay-as-you-go model is expensive compared to a subscription model, but I don't want to marry to a single provider.

Thanks for bringing OpenCode Go to my attention. Your comparison is the research I didn't know I needed, and I will be cancelling my Copilot subscription to replace it with OpenCode Go right away.

It's clear to me that the maintainer is referring to "shushtain" and those type of people

when they take that tone with you.

This makes it sound as if you took it personally?

Having a bad day does not entitle you to take it out on others

You should totally post this on the original thread just for adjustment :-)

I am definitely looking forward to TurboQuant. Makes me feel like my current setup is an investment that could pay over time. Imagine being able to run models like MiniMax M2.5 locally at Q4 levels. That would be swell.

Not the answer that you are looking for, but I am a fellow AMD GPU owner, so I want to share my experience.

I have a 9070 XT, which has 16GB of VRAM. My understanding from reading around a bunch of forums is that the smallest quant you want to go with is Q4. Below that, the compression starts hurting the results quite a lot, especially for agentic coding. The model might eventually start missing brackets, quotes, etc.

I tried various AI + VRAM calculators but nothing was as on the point as Huggingface's built-in functionality. You simply sign up and configure in the settings [1] which GPU you have, so that when you visit a model page, you immediately see which of the quants fits in your card.

From the open source models out there, Qwen3.5 is the best right now. unsloth produces nice quants for it and even provides guidelines [2] on how to run them locally.

The 6-bit version of Qwen3.5 9B would fit nicely in your 6700 XT, but at 9B parameters, it probably isn't as smart as you would expect it to run.

Which model have you tried locally? Also, out of curiosity, what is your host configuration?

[1] https://huggingface.co/settings/local-apps [2]: https://unsloth.ai/docs/models/qwen3.5

What is this 10€ per month subscription that you are talking about?

I don't understand why I'm getting downvoted.

I am legitimately curious about the parameters that the person used for running the model locally to get the results they got because I am myself currently experimenting with running models locally myself. You can see I am asking similar questions to others in this same thread and correlate the timestamps.

At what temperature did you run it and what was your context limit?

Apparently there is a whole science behind running models. I have seen the instructions that unsloth publishes for their quants and depending on the model they'll tweak things like the temperature, top k, etc.

The size of the quantization you chose also makes a difference.

The GPU driver also plays an important role.

What was your approach? What software did you use to run the models?

What front-end framework did you use? I find the UI so visually appealing

Which quantization are you running and what context size? 32tok/s for that model on that card sounds pretty good to me!

It might be that the system prompt sent by codex is not optimal for that model. Try with open code and see if your results improve

By anyone do you mean a well-established business or any entity willing to serve you?

[...] _but not necessarily use the right format._

This has also been my experience. But isn't the harness sending the instructions on how to invoke a tool? Maybe it is missing the formatting part. What do you think?

Through my Kagi subscription I get access to quite a few models [1] but I tend to rely on Qwen3 (fast) for quick questions and Qwen3 (reasoning) when I want a more structured approach, for example, when I am researching a topic.

I have tried the same approach with Kimi K2.5 and GLM 5 but I keep going back fo Qwen3.

I also have access to Perplexity which is quite decent to be honest, but I prefer to keep everything in Kagi.

1: https://help.kagi.com/kagi/ai/assistant.html#available-llms

Great idea of inferbench (similar to geekbench, etc.) but as of the time of writing, it's got only 83 submissions, which is underwhelming.

[...] it's much easier to fine-tune a "general" model into performing some very specific custom task (like classifying text, or translation, etc)

Is this fine-tunning process similar to training models? As in, do you need exhaustive resources? Or can this be done (realistically) on a consumer-grade GPU?

But are we really at the point yet where people are running local models without knowing what they are running them on..?

I can only speak for myself: it can be daunting for a beginner to figure out which model fits your GPU, as the model size in GB doesn't directly translate to your GPU's VRAM capacity.

There is value in learning what fits and runs on your system, but that's a different discussion.

Apparently there are a few more similar communities like the one from the post

https://tildeverse.org/members/

Pi ships with powerful defaults but skips features like sub-agents and plan mode

Does anyone have an idea as to why this would be a feature? don't you want to have a discussion with your agent to iron out the details before moving onto the implementation (build) phase?

In any case, looks cool :)

EDIT 1: Formatting EDIT 2: Thanks everyone for your input. I was not aware of the extensibility model that pi had in mind or that you can also iterate your plan on a PLAN.md file. Very interesting approach. I'll have a look and give it a go.

I agree with you, especially with this:

They paid for the access the same as any other.

If anything, this makes them more legit than Anthropic because they are paying for the content, whereas Anthropic just stole *all* the data they got a hold of. So, in this case the Chinese AI labs stand on higher moral ground LOL.

The article touches a bit on how Sega basically lost. There is literally a whole documentary about this: Console Wars, where they go deep into how Sega lost the battle: https://en.wikipedia.org/wiki/Console_Wars_(film)

Hello. I am happy to take this for a spin.

I see that not all models available in my Github subscription are available (all models should be visible).

Further, is it possible to use openrouter with the current implementation? I couldn't figure it out by reading the documentation alone.

Thank you!

This is definitely a cool finding.

Have you investigated more on this topic? like, anything similar in concept that competes with Serena? if so, have you tested it/them? what are your thoughts?

Curious question: why would they check for installed extensions on one's browser?

I had the dilemma of choosing between Bazzite and CachyOS or Manjaro for my workstation, which is also my gaming rig.

The whole immutable distro felt like a hinder for my workflow (running docker containers, etc.), so that's why I went for CachyOS.

I ran Archlinux as my main driver on both PC and Laptop for more than a decade but after having the opportunity to use a Windows machine with WSL and eventually WSL2, I felt like I had access to the best of both worlds: a Linux terminal for development (bash + tmux + vim, now bash + zellij + neovim) without the hassle of updates breaking things every few months and a out-of-the-box native gaming experience.

But with the enshitification of Windows (first all the spam and ads on the Start menu, then Microsoft forcing you to have an account to be able to use the machine and the expensive license for Windows Professional if you want access to Hyper-V, which I did), I did some research, tried a few new distros (Manjaro, Bazzite and CachyOS) and settled for CachyOS (gaming support was the main driver, based on Archlinux was secondary).

I do everything I did on Windows and some more: all the terminal stuff plus browsing, CAD modeling, 3D printing / slicing, Office stuff... I miss nothing. No more double partition to boot into Windows when I want to game.

My RX 9070 XT runs smoothly with no driver issues whatsoever. I even have tested the waters running some LLMs with LM Studio and that also worked out of the box.

The only thing that has been a bit meh are Teams and Slack and I believe that has to do with the fact that I ran them in Firefox. Once I ran Slack on Chromium, noise canceling was again available.

2009 was the year of Linux on desktop for me. 17 years later, after going back and forth between macOS and Windows, it feels good to be back home.

One last note in my random ramble is that I do not have as much spare time as before, and I had heard this from other people back in the day whenever I'd say I ran Archlinux on my machines, so I am going to repeat what others have said to me: it's really nice to not have to worry about much, be able to sit down and get productive right away. To me, CachyOS and KDE have made that idea my actual experience and for that I am grateful.