Bloated Test Directory

Founder of a Useless tests and CI directory that runs for way to long

[Bloated Test Directory @OpenSourceSlop]

@fluffypony @navtechai @itsPaulAi A good analogy is “putting on a button up shirt, where the first button that you do is in the wrong slot”

It’s fine if you don’t need accuracy, but if accuracy is important to you then even the cheaper option doesn’t make sense.

[Bloated Test Directory @OpenSourceSlop]

@h14hdotcom @tmaiaroto @loktar00 Yea that’s what perplexity is trying to do now

[Bloated Test Directory @OpenSourceSlop]

@mohittwwt The unified memory is the real bottleneck. A year ago if i had the opportunity to get 1.2 T in unified memory I would have jumped at the chance.

Now with OSS models needing that plus more you would need 2 or 3 and you will likely be behind the behind in 6 months 😅

[Bloated Test Directory @OpenSourceSlop]

With 2 you could run Kimi K 3 at home locally. It’s only $40,000 😅

[Marques Brownlee @MKBHD]

M5 Ultra has also landed

- Up to 36-core CPU
- Up to 80-core GPU
- 1.2TB/s for unified memory

(A fully specced-out M5 Ultra Mac Studio with 16TB storage and 256GB of memory now costs $18,299)

https://t.co/h2XOgf5P82

[Bloated Test Directory @OpenSourceSlop]

@LucaCaponeX @LinusEkenstam This 👆

[Bloated Test Directory @OpenSourceSlop]

@RujiraNetwork @Vultisquad I would love to see a deeper integration with Vultisig.
I understand the desire for sonar but Vultisig integration would be a better fit in my opinion
No need to spend time on another wallet. Just bring Vultisig into the fold

Win Win 🏆

[Bloated Test Directory @OpenSourceSlop]

@LLMJunky Glad I got one before prices went up.

[Bloated Test Directory @OpenSourceSlop]

@LLMJunky This hit me right in my feels 🥲

[Bloated Test Directory @OpenSourceSlop]

@mohittwwt Sounds like ultra fast is going to be released soon.

You already know the play

[Bloated Test Directory @OpenSourceSlop]

@melvindvivas @federicodonaton It’s because outside of Terra High you can get better intelligence for about the same price using Sol. Anything less then Terra High and you might as well roll the dice and use Luna Max for a fraction of the price

[Bloated Test Directory @OpenSourceSlop]

This is worrying because they make the best Frontier models for actual work currently IMHO.

If all their top people are leaving its because they don't want to be left holding the "bag"

[Coin Bureau @coinbureau]

🚨SHOCKING: All major OpenAI executives QUIT, leaving Sam Altman to steer the company, as revenue tops $40 BILLION.

OpenAI’s Chief Revenue Officer just quit after only 8 months, while its COO also departed yesterday even as revenue surges ahead of a potential IPO.

The company’s head of ethics, head of safety systems, and former head of mission alignment have ALL exited in recent weeks. At this point, there may be no major executive left to quit besides Sam Altman.

[Bloated Test Directory @OpenSourceSlop]

@Tukytuky_ @FabianoSolana I use @vultisig since it gives me fully customizable protection levels for my different wallets

[Bloated Test Directory @OpenSourceSlop]

@LLMJunky It just needs to be good enough.

And they have that already. And it’s shipping on every IPhone and Android starting EOY.

Personal context is why they will win

Most people will happily give google all of their personal data in exchange for a personal on device assistant

[Bloated Test Directory @OpenSourceSlop]

@mohittwwt @scaling01 I don't know why but these limit resets are almost always when I have +90% of my weekly limit left LOL

[Google DeepMind @GoogleDeepMind]

Gemini 3.7 Flash is here.

It’s stronger for coding, knowledge work, and web development. 🧵

[Bloated Test Directory @OpenSourceSlop]

@mohittwwt @pejmanjohn @ajambrosino Im not 100% sure but i know that they market it as a better optimized way to interact with Codex, Claude etc...

In the demos I've seen they are always clicking around and it seems very responsive

[Bloated Test Directory @OpenSourceSlop]

@mohittwwt @scaling01 Honestly I don't think its worth the cost for the average solo dev to use. Or I should say its not worth the cost to me. I currently don't use fast mode anyway because of cost so why would I use the more expensive "faster mode" even if i think its gonna be magical

[Bloated Test Directory @OpenSourceSlop]

@mohittwwt @pejmanjohn @ajambrosino Yea this is the hardest part of using this. Its the main reason i keep thinking of switching to T3 code

[Bloated Test Directory @OpenSourceSlop]

750 TPS with voice must feel like magic

[OpenAI @OpenAI]

Powered by @Cerebras, Ultrafast generates up to 750 tokens per second, bringing our most intelligent model to products and workflows where every second counts.

Ultrafast is designed for businesses where faster frontier intelligence creates a measurable advantage, including real-time voice and customer support, commerce, coding and design, financial research, and security response.

https://t.co/QLBhmodeI4

[Bloated Test Directory @OpenSourceSlop]

Grok 4.6 medium scores Higher on DeepSWE v1.1 then high does.

Honestly its comparable to Codex Sol models and if your on API it might make sense but on a sub i have yet to see a model that I would replace Codex Sol with.

[SpaceXAI @SpaceXAI]

Introducing Grok 4.6.

It delivers frontier intelligence and is a significant improvement over Grok 4.5 at the same price.