Busonix

[Busonix @busonix]

@sama can't we just all be friends

[Busonix @busonix]

I think Nvidia obviously has incentive for local ai to thrive, so they can sell more equipment and now train up their own models from all the HF data. But, I still see them complying and banning certain models deemed a threat by the government. I don't think the average Joe dipping their toes into local ai in the future will be hunting uncensored models. It's just us few nerds that are into that

[Busonix @busonix]

@TimJayas That and I assume the data xAI gets is the main reason

[Busonix @busonix]

@thegenioo Yeah not too surprising honestly, lots of juicy data they don't want xAI to have

[Busonix @busonix]

@kimmonismus This is the most painful part of the 5 hour limits🙃

[Busonix @busonix]

(5/5) Even More Value

- YouTube Premium Lite ($7.99/mo value)
-Google Health Premium ($9.99/mo value)
-Google Home Premium Standard ($10/mo value)
-Access to Gemini Spark an in the cloud agent
-Google Suit with Gemini built in (Docs/Slides/Sheets etc..)
-Expanded access to image, music, and video generation models in Gemini, Search, and Google Flow

[Busonix @busonix]

$20 a month for upwards of $$$ in value

Most people don't even consider Google's AI Pro plan or are subscribed and barely use what's available.

Here's what you get and why it's actually worth it 👇

[Busonix @busonix]

(4/5) Gemini Notebook (NotebookLM)

The perfect tool for students and researchers. You can add up to 300 sources which then will be queried against when you ask questions. This keeps research and studying grounded in trusted sources you add.

It also will generate artifacts based off your sources:
-Audio/Video Overviews
-Slide Decks
-Flashcards
-Mind Maps
-Reports and more

[Busonix @busonix]

With local models there’s no need to worry about this happening!

[Tibo @thsottiaux]

We unfortunately have decided that we cannot continue providing access to our models through Cursor and are ending our partnership. It boils down to trust and we’ve asked that this takes effect on November 12 to give you some time to plan.

Many have used the GPT models through Cursor and here are options we know should work in the future:

- We will continue to allow using your own OpenAI API key and similarly will continue to provide access through our IDE extensions for Cursor.
- We will keep working with the broadest range of tools and harnesses, some of which are OSS, but also many many closed-source ones.

We are as committed as ever to continue supporting developers and the flourishing ecosystem of tools, harnesses and products. We will also continue to invest in our own open-source initiatives and believe in broad optionality for developers.

You can read more about our decision in the blog:
https://t.co/Oj76Bkc2KX

[Busonix @busonix]

Anyone else feel like Opus is dumber lately? And unnecessarily difficult to when trying to get it to do something? Driving me nuts

[Busonix @busonix]

I know most people will obviously recommend the 256gb studio, more memory is better yeah.. but at the price point for the 128gb studio, I think it’s a perfect build for most people who can afford it.

With MOE and SSD streaming, it can punch above its weight

[Busonix @busonix]

It’s easy to hate on Google, but I generally have a good experience with using antigravity and Gemini flash

[Busonix @busonix]

@jun_song And your local models only get better, and never disappear 🙂

[Busonix @busonix]

@jun_song I think regardless some regulation is coming, I’m sure Nvidia has no hesitation in cutting out some competition’s models from HF. They get to offload the blame to the government. But that won’t stop people getting hands on the models they want, thankfully

[Busonix @busonix]

There’s clearly a massive gap between open source and closed source when it comes to the overall inference setup and product.

That’s why local feels dumb even with newer and improved model releases.

[BridgeMind @bridgemindai]

GLM 5.3 Flash and Qwen 3.8 Flash are impressive for their size. But something is missing.

They benchmark well. They run locally. They cost almost nothing. And then you use them for real work and you feel the gap immediately. The polish is not there. The reliability is not there.

Open source AI running locally is genuinely good for the world. But if you have a real budget, there is no spot in your workflow where these fit. Fable 5, GPT 5.6 Sol, and Grok 4.6 are better at everything that matters.

Local models are for people with time. Frontier models are for people with deadlines.

[Busonix @busonix]

@jun_song Next deepseek model’s gonna hit like crack

[Busonix @busonix]

@TencentHunyuan Flash when?? 🙂