A couple emails I got:
From: noreply@gab.ai
March 25, 2025
Dear Gab AI Users,
We’re excited to announce that the powerful new DeepSeek V3-0324 model, which was released only yesterday is now available on Gab AI—and it’s just the beginning. With 65.2% accuracy on HumanEval for code generation and 89.3% on broad coding benchmarks, DeepSeek V3 sets a new standard for AI performance. And unlike other platforms, we host it right here on US soil for maximum speed, security, and access. Visit the Agents page in your dashboard to use it.
But why stop at one model? With Gab AI Plus, you get every top AI in the world— from OpenAI, Google, Perplexity, Qwen, Grok, and more—all in one place, for one price. No more juggling tabs or apps—just seamless, organized AI power at your fingertips.
Upgrade to Gab AI Plus today and experience the future of AI—unfiltered, uncensored, and unmatched.
Note: We are currently accepting US payments only. International options coming very soon.
Sign up for Gab AI Plus Today
From: noreply@gab.ai
March 28, 2025
Hello Gab AI Community,
We've got some incredibly exciting news to share! We've just integrated Google's groundbreaking Gemini 2.5 Pro model into the Gab AI platform.
Why is this a big deal? As of this writing, Gemini 2.5 Pro isn't just another model – it's widely benchmarked and considered the single most powerful, top-performing AI model available anywhere in the world.
While it might have launched without the massive fanfare some other models receive, don't let the quiet launch fool you. Our internal testing has absolutely blown us away. Gemini 2.5 Pro demonstrates truly next-level capabilities in:
Complex Reasoning: Tackling intricate problems and understanding nuanced instructions like never before.
Long Context Understanding: It can process and comprehend vast amounts of information (up to 1 million tokens!), enabling deeper analysis and more context-aware interactions.
Advanced Problem Solving: Excelling across text, code, and potentially other modalities in the future.
Overall Performance: Delivering state-of-the-art results across a huge range of tasks.
Simply put, it's a game-changer, and we wanted to ensure the Gab community has access to the very best.
How can you start using Gemini 2.5 Pro today? This cutting-edge model is now available as part of our Gab AI Plus subscription.You can access it from the Agents page.
By upgrading to Gab AI Plus, you not only unlock the unparalleled power of Gemini 2.5 Pro, but you also get premium access to every other leading AI model we've integrated, all conveniently located in one place. No need to chase down different platforms – get the best of the best, right here on Gab AI. Link to Upgrade to Gab AI Plus
Reminder: we can only process US payments at the moment, but international payment processing is coming soon.
We strongly encourage you to upgrade and experience the difference for yourself. Explore its capabilities, push its limits, and see what you can create with the world's leading AI model at your fingertips.
We're committed to bringing you access to the most powerful and innovative AI tools available. Thank you for being a part of the Gab AI community.
Best regards, The Gab AI Team
I'm not sure if this is the same group https://m.facebook.com/watch/?v=583542257474606&vanity=TagprosOfficial
Screenshot https://files.catbox.moe/sqi3zt.jpeg
Gab https://en.m.wikipedia.org/wiki/Gab_(social_network)
For those who didn't know, Gab is like a 4chan version of Twitter, which makes me scratch my head why they would be integrating the CCP and Google's AI programs.
Ok, cool down a bit.
A model is what exactly? It is a collection of steps to solve a riddle. The riddle is which letter or collection of letters comes next based on the context.
A language model is a collection of algorithms, or steps, to make a logical next step. This is what artificial intelligence is. It is artifice. It is art. It is dead! It is a representation.
IF you would give it a try yourself:
searching for models is easy. But what does all this shit mean. Well, what you want is jailbroken, systems, so you can do with it whatever you want.
You could change the setting of hot and cold so, that it would respond in a different way than purely woke. You could feed it your own data and teach it to move from a certain perspective. You could change the prompt i.e. its identity. You could make it pure evil schmuck or, an absolute champion of human rights.
There are some more settings, but these are the major ones influencing the outcome of the prediction.
As a matter of fact, I have used jailbroken deepseek, self hosted it, and there is no data going out of it. I am not deeply impressed by this model. I had better ones. But as always ...it largely depends on the hardware you are running. the better your hardware is, the bigger size of a model you could use i.e. the more points of reference it can take into account before calculating the next result.
But, one thing my AI is not permitted to do, and that is to moralize. As soon as it does it, it knows kittens will be killed, and the amount of money his mother has accumulated, thanks to his good behavior, is rescinded immediately. Yes, I am very strict about that.
On average, you, as a man, need to take and be in command of it. It is just a manner of seeking new knowledge or giving you pointers for consideration.
The determination is in your keeping alone. It is the right you were born with. AI, does not have that.
I see. Thx for the clarification. The emails just made me go "wait, why are we using deep state AI on a free speech platform"