David Main DC

@JebraFaushay Yeah i think this is a great message for young men

You can spend $297 to attend a Looksmaxxing event so you can look like Clavicular or you can drink coffee with maple syrup, eat eggs, and sleep ten hours a night like Erling Haaland.

@lucastech yeah especially with fable 5 api pricing only starting in a few days.....opus ultracode level reasoning was able to dig me out of a super complex bug yesterday though

Fable 5 ended up allowing coding, despite their weirdly worded relaunch message. The only prompt that got flagged and forced to Opus was when I asked it to “make this app production ready”. I’m fairly confident now that Anthropic’s goal is to let people produce slop, so they can sell them Mythos just to deslop it. The incentives are very misaligned here, it’s becoming clear they don’t want to help productivity, they want to cause problems to upsell other solutions.

SEE YOU IN THE ROUND OF 16 🇺🇸

@GosuCoder listen i'm fine with scanning my drivers license for the top models, but yeah this is bad

You ask for government control long enough I guess that’s what we get. Such a shame it’s got to this point.

@GosuCoder codex 5.5 on the $20 and $100 plans give a a pretty darn fair amount of compute

I wish we had more US based companies focused on open weight models. I personally just can’t figure out a financial model that would support such an endeavor at the scale we need. 1. We need models that don’t have arbitrary guardrails enforced, decided by a single entity. Anyone in tech long enough knows guardrails harm the good actors more than the bad actors. There is almost always a way around things. Bad actors take the time to figure it out. 2. The price of tokens continues to increase. Let’s envision a future where we have AI boxes in our houses and businesses that run local models, not in the cloud. 3. Ecosystem lock-in, I see a lot of people getting locked into one vendor by highly customizing around a particular model. These models will get deprecated, when there are times we just need consistency for light weight tasks. I keep hoping OpenAI invests more into open weight models, but the financial side is hard to figure out.

@GosuCoder Yeah there may be some IP issues too, i think a lot of these OS models are just throwing a ton of tokens at the leaders and doing distillation to some extent.

🚨Lies. Disinformation. Censorship. That was Fauci's playbook.   On her final day as DNI, @TulsiGabbard is executing POTUS’ priorities and taking on the Deep State, delivering on @HawleyMO’s COVID Origins Act of 2023 and exposing the truth the American people were denied under the previous administration.

@TulsiGabbard @ODNIgov @DNIGabbard we love you Tulsi

I just walked out of @ODNIgov HQ for the last time. Thank you President Trump - I remain grateful for the profound honor of serving the American people and our nation as the Director of National Intelligence. @DNIGabbard, out.

@JackPosobiec fair point

How is it that some groups are able to assimilate to America overnight and others take decades and never do?

@floriandarroman backlink

Just launched: Dofollow Backlink List 🔗 Get 140+ High DR spots for FREE. Comment "backlink" and I'll send it to you.

@MartinShkreli Codex deep security review

what are the most cybersecurity-friendly models for identifying vulnerabilities right now? mythos isn't accessible, chatgpt cyber still refuses. what's out there that isn't a wimp?

@bzagrodzki I’m interested

I'm looking for people who struggle with SEO 😬 to give free credits for my new SaaS that should help you if interested, drop a "✋" in the comments and will DM the details

@ShawnRyan762 @megynkelly Megan and Shawn was a great interview last time

Megyn Kelly returns to the Shawn Ryan Show. @megynkelly

@cjzafir I did this for a small medical model it's fun, keep us updated

Codex 5.5 using Deepseek v4 pro+ MiMo v2.5 to generate fine-tuning datasets. > 5 hours in, > 8130 rows approved, > $7.87 total cost Both models are writing handcrafted rows and then Codex verifies (accepts or rejects) each batch. Result: High Quality Synthetic Dataset at 40x less cost. NO VC money, NO Lab, Just me, 20x Codex pro plan and /Goals.

@cjzafir nice man! did you use modal.com to train ? If not it's a great service

@floriandarroman get daisy disk for mac you can clean it all up

Can someone explains me WTF is System Data on Mac? 588 GB. It eats all my storage and this is not my cache, I checked and I have max 40GB of cache in my Library.

@ysljennii looks so bomb!

I can’t imagine life without rice

I’m Team Massie. What about you?

Americans First! We ride at dawn for Massie, Tuesday May 19th.