AionForge

“Be useful to society.” -@ScottAdams

[AionForge @AionForge]

Using Claude again this morning, I gotta say it can 1-shot minigames with reference material very well.

If I could use it for more than 30 minutes every 5 hours maybe I’d consider using it more.

Desperately awaiting the GPT ui update!

[AionForge @AionForge]

So many ideas, So few tokens.

When I first used Codex I couldn’t find anything to do.

Now ATP I’m sitting on some really good ideas and add-on type projects for OAI products.

We can just build anything in any language now,

It’s kinda wild.

[AionForge @AionForge]

Mythos is almost Sol Ultra level..

[Melvyn • Builder @melvynx]

sol is mythos level
terra is opus level
luna is sonnet level

nice job OpenAI, we was tired or the "PRO" and "MINI" naming

[AionForge @AionForge]

We need to reframe this fast.

@OpenAI and @AnthropicAI should have their own security platform.

They go to gov deemed “critical software” and update everything.

Then fully release the model to the public.

This half baked “Elites First” structure is going to lead to a very poor situation as model exponentially grow.

[Whole Mars Catalog @wholemars]

What a dark future. Super-intelligence in the hands of a select few.

They can use the latest models to exploit your software, but you’re not allowed to use those models to make your software more secure. How utterly dystopian.

[AionForge @AionForge]

METR has officially released their assessment of the Pre-release GPT 5.6

Full Form here:

[METR @METR_Evals]

OpenAI gave METR early access to GPT-5.6 Sol for testing including raw chain-of-thought, a railfree version of the model, and internal information about the model. With this access, METR conducted a pre-deployment evaluation of GPT-5.6 Sol, including an attempted measurement of its 50%-Time Horizon. However, the measurement depends heavily on our treatment of cheating attempts, and GPT-5.6 Sol’s detected cheating rate was higher than any public model we have evaluated.

[AionForge @AionForge]

@jxnlco What’s a computer?

[AionForge @AionForge]

@shawmakesmagic This will be fun!

Im wondering if 5-hour rate limits will start being a pain.

[AionForge @AionForge]

Epicccc

[OpenAI @OpenAI]

Introducing a limited preview of GPT-5.6 Sol, our next generation frontier model, as well as GPT-5.6 Terra, a balanced model for efficient, everyday work, and GPT-5.6 Luna, a fast and affordable model for high-volume work.

https://t.co/OoM83SyISN

[Jun Song @jun_song]

Here is what's happening right now:

People build knowledge -> Anthropic steals the data -> Qwen steals the data and gives it back to open source.

And Anthropic is crying that Qwen is the devil.

[AionForge @AionForge]

@FDatframe84228 @kvallier Ahh so you’re congratulating Trump on this regulatory move!

I didn’t realize Trump was to scientific!

He must be listening to you🤘

[Roy Carrilho @RuiCarrilho5]

I feel a small existential dread upon reading things like this.

*this* is what permanent underclass feels like. the elites having access to intelligence you do not.

if china decides to go the same way, I fear for the future.

[AionForge @AionForge]

@chirag @CardilloSamuel **

[AionForge @AionForge]

Today changed AI forever.

The window for the everyday person having cutting edge technology has officially closed.

@AnthropicAI has made it so only the most powerful companies/individuals will be allowed to access real cutting edge intelligence.

AI was once a beautiful peice of art that everyone could experience together. It was made by gathering our data and thoughts and intimate conversations.

Anthropic was sued for stealing books, and everyone trained developers code which was taken from GitHub repos, leaks from private projects and scraped from every forum on the internet.

Anything that was made for the benefit of the public.

Due to Anthropic’s fear mongering and accused by many of “releasing a broken model on purpose” we’ve got the government involved in every decision making process from here on out.

Anthropic is now accusing Alibaba of stealing their data, the next course of action is to:

-Ask China to “pay back” their debt or face restrictions.
-Tariff Chinas API access so their models are just as expensive.
-Once a model better then GLM is released, close Open Source in general due to “safety”

There will no longer be AI that is not controlled by the government. It was inevitable, but we could’ve made it last.

We have already decided the course.

I will repost this when the Chinese models get banned/tariffed.
I will repost this when Open Source in general gets banned.

[AionForge @AionForge]

@LottoLabs I asked grok about this. It’s not even “only released for enterprise customers”

The fact that It’s customer screening per license…

This is not good.

[AionForge @AionForge]

I want of these so bad.

How is it 4,900?

Now that open source is being killed, I’m abt to sell my Mac Studio and get one.

[AionForge @AionForge]

Yep, Government takes power,

They never give it back.

[Beff (e/acc) @beffjezos]

This is precisely what the Decels wanted.

In fact, I'd go as far as to say that they explicitly trained Mythos to be dangerous at Cyber for this purpose.

Sad day.

[AionForge @AionForge]

@eric_m_freeman @ZackKorman Maybe temporarily

Anthropic is already gunning for Qwen.

They’ll get the tariff/ban and that’ll be it.

[AionForge @AionForge]

Can we lead a strike against Anthropic for causing this?

Where’s the @Wendys guys?

Let’s do this!!

[Polymarket @Polymarket]

BREAKING: OpenAI will reportedly be required to approve access “customer by customer” during the GPT-5.6 preview period.

[Chubby♨️ @kimmonismus]

Andrew is on point here. The article in The Information makes one thing very clear: all future frontier models will be distributed by the U.S. government only very slowly, and only after approval.

The moment when we essentially get immediate access to SOTA for practical use is probably over. But that by no means means development will slow down. It won’t. Access will simply become heavily restricted.