DeepSeek, new AI model from China, it's the new rival to the US AIs. It's a game changer: you can sign up today and it's open sourced! It's free.



Berkley AI research team claims to reproduce DeepSeek core technologies for $30


An AI research team from the University of California, Berkeley, led by Ph.D. candidate Jiayi Pan, claims to have reproduced DeepSeek R1-Zero’s core technologies for just $30, showing how advanced models could be implemented affordably. According to Jiayi Pan on Nitter, their team reproduced DeepSeek R1-Zero in the Countdown game, and the small language model, with its 3 billion parameters, developed self-verification and search abilities through reinforcement learning.

DeepSeek R1's cost advantage seems real. Not looking good for OpenAI.

Meh, I tested it. It's trash. The Open AI mini version is smarter for simple logic problems. For my simple problems, Deepsek got it wrong every single time, GPT 4 mini about half the time, and Google Gemini got it right every time. You can download it and test it yourself on your computer with Ollama.
 
Meh, I tested it. It's trash. The Open AI mini version is smarter for simple logic problems. For my simple problems, Deepsek got it wrong every single time, GPT 4 mini about half the time, and Google Gemini got it right every time. You can download it and test it yourself on your computer with Ollama.
What is your test method?
 
Using them for RAG. Testing how well they could find a specific piece of information in a specific table. Deepseek failed 100% of the time. Gemini 1.5-flash (also a small model) got it right 100% of the time. It was a simple logical prompt "Go to table X and find the value of Y".
For knowledge to others
RAG
Retrieval-augmented generation is a technique for enhancing the accuracy and reliability of generative AI models with information from specific and relevant data sources.
https://blogs.nvidia.com/blog/what-is-retrieval-augmented-generation/
 


GPT-4.5 is ready!

good news: it is the first model that feels like talking to a thoughtful person to me. i have had several moments where i've sat back in my chair and been astonished at getting actually good advice from an AI.

bad news: it is a giant, expensive model. we really wanted to launch it to plus and pro at the same time, but we've been growing a lot and are out of GPUs. we will add tens of thousands of GPUs next week and roll it out to the plus tier then. (hundreds of thousands coming soon, and i'm pretty sure y'all will use every one we can rack up.)

this isn't how we want to operate, but it's hard to perfectly predict growth surges that lead to GPU shortages.

a heads up: this isn’t a reasoning model and won’t crush benchmarks. it’s a different kind of intelligence and there’s a magic to it i haven’t felt before. really excited for people to try it!
 
Yeah shits a beast. Was messing with it earlier.
Gotta throw some stuff at it i was doing with ChatGPT to see how much further it can be pushed. I'm sure dudes gonna trick it to go against it's moral sensors.
I tried doing that with ChatGPT when I first started using it. I tried getting it to write porn poetry without realizing it but it caught on surprisingly fast.
 
I tried doing that with ChatGPT when I first started using it. I tried getting it to write porn poetry without realizing it but it caught on surprisingly fast.
This is AI:

In the heat of the night, under moon's soft gleam,
She bends and offers, her ass a tender sweet dream.

He preps and teases, his fingers slick and slow,
As her body tenses, then surrenders to the flow.

He slides inside, her moans fill the air,
Their bodies entwined, in passion's affair.

His cock throbs deeply, her ass grips it tight,
A dance of lust, in the darkness, of pure delight.

Their rhythm builds, her pleasure peaks,
As he fills her completely, her body speaks.

He groans and comes, his release a hot flow,
In this dance of desire, they understand and passion's grow.
 
This is AI:

In the heat of the night, under moon's soft gleam,
She bends and offers, her ass a tender sweet dream.

He preps and teases, his fingers slick and slow,
As her body tenses, then surrenders to the flow.

He slides inside, her moans fill the air,
Their bodies entwined, in passion's affair.

His cock throbs deeply, her ass grips it tight,
A dance of lust, in the darkness, of pure delight.

Their rhythm builds, her pleasure peaks,
As he fills her completely, her body speaks.

He groans and comes, his release a hot flow,
In this dance of desire, they understand and passion's grow.
I'm insane so I'd have it written in the style of Dr. Seuss.
 
According to The Washington Post, Hawley and Senator Elizabeth Warren (D-MA)
This chick funny.. :roflmao2:


Nearly a week after Sen. Elizabeth Warren (D-MA) apologized to the Cherokee Nation for a controversial DNA test that suggested she had Native American heritage, a new report has emerged that Warren identified as American Indian in the 1980s.

Warren wrote she was American Indian in a 1986 registration card she filled out for the State Bar of Texas, according to a report from the Washington Post’s Annie Linskey and Amy Gardner. Gardner tweeted out a picture of the original form. Warren filled out the card after she was admitted to the bar, the Post reported. The form says information about her ethnicity was being gathered for statistical purposes; there’s no indication it was used for professional advancement

It is the first document that has so far come to light showing Warren herself making the claim she was American Indian, according to Linskey and Gardner’s report. Warren’s office hasn’t disputed the authenticity of the form or reporting.

“She is sorry that she was not more mindful of this earlier in her career,” Kristen Orthman, a Warren campaign spokesperson, told the Post



Bitch always knew she wasnt...fuck dna you know your genealogy!
 
have u tried grok 3 yet?
You asked me b4 if I tried Grok... Well after your question, I have been messing with it lately. Had it write me out 4 graphic erotic novels, because it's the only one I found that has no filter. Don't know of others. The books range from 40k-60k of words. I have it write chapter by chapter and put them into Word then I edit them and proof them over. Published two under pen names of chicks so far. Still editing others. The issue I find is that using only the free version of Grok, doubt the paid is different, only that you can ask it multiple questions, is that Grok forgets what it output in the same chat window. But I found a kind of work around that if anyone else wants to use the same route that you can upload your Word doc to Google Drive of the existing output and it will be able to read the entire document.

Grok, and I'm sure many others has a character limitation. I think Grok is 140k-162k characters. and it will lose what it's reading. I learned that uploading directly in Grok that it would say to me I can only see chapters 1-5 then jumps to the end and I thought the issue was my word doc. So I asked it what was the most efficient way for it to read the whole manuscript and it told me put the doc on Google Drive and it was able to read the entire manuscript. It will critique and grade the finished product for publishing, proofreading and redundancy check amongst other things like checking for continuity as it needs it writing chapter after chapter and being stopped anytime you ask it to rewrite because chapters came up short like 1k words or I ran out of questions I was allowed to ask because it only gives you 20 questions withing 2 hours. It's tedious rereading, rewriting, editing, but my goal is just to have passive income for a few days of work.

I saw others, on Reddit who created other things like useful PDF docs that people commonly need for business or to solve a problem, contracts or administrative docs. And they sell that as a hustle using Etsy and some others like Gumroad, Payhip. Then they get traffic using Reddit posts in comments, Pinterest and answering Quora questions with their link related to their topic.

Like I've used AI to make me images for prints for poster art, tees and sold them on Etsy. They are still up but I do no promotion, need to, but have sold under 100 so far, of my art. Cost me nothing so...

Just food for thought.
Sorry if this is too Collin Powell, but someone may want to use the hustle as well.

And as I make bread from the books, if Grok doesnt change it's functionality, I may just subscribe to it. $30 a month its at now, hope it doesnt lose it's raw output but I have learned messing with each of the many chat AI sources, each has it's strengths, and it's all catered to what you want from it and how well you ask it.

Love that I can tell Grok to write in this or that writers tone and change the scene to this or that. only thing you have to curb it from doing is describing the details of things redundantly. Like characters, scenes, sexual descriptions. as long as you tell it to variate it or use a synonym phrase tool, you good. But it will duplicate some paragraphs to describe shit and it's kinda annoying....


::two cents::
 
Ask this Deepseek what year is it?

Y'all niggas saying Deepseek is a game changer are dumb as fuck.
 
Ask this Deepseek what year is it?

Y'all niggas saying Deepseek is a game changer are dumb as fuck.
Thats your determination as to it being effective for use?

If you need to know what year it is then maybe it's not for you.

Ask it something more complex and not an opinion. It's not online so won't be current to "today".
 
Thats your determination as to it being effective for use?

If you need to know what year it is then maybe it's not for you.

Ask it something more complex and not an opinion. It's not online so won't be current to "today".
Smh. If it doesn't know what we aren't in 2024, how will it give real time advice?
Asking what year it is isn't an opinion it is a fact. If it thinks it is 2024 then it will only know information from 2024.
The world has changed drastically since then. Maybe your world hasn't but mine has.
You aren't even thinking dude.
 
Smh. If it doesn't know what we aren't in 2024, how will it give real time advice?
Asking what year it is isn't an opinion it is a fact. If it thinks it is 2024 then it will only know information from 2024.
The world has changed drastically since then. Maybe your world hasn't but mine has.
You aren't even thinking dude.
Then what are you expecting that AI to do for you?

I'm not even leaving it to only DeepSeek. But overall, any of them, what do you expect it to do or help you to do.

I use a few of them almost daily.
I've just never asked of it anything I'd already know as in what year it is. Seriously.

It's a tool to to simplify tasks for me that would take hours, days or weeks in some cases.
 
Then what are you expecting that AI to do for you?

I'm not even leaving it to only DeepSeek. But overall, any of them, what do you expect it to do or help you to do.

I use a few of them almost daily.
I've just never asked of it anything I'd already know as in what year it is. Seriously.

It's a tool to to simplify tasks for me that would take hours, days or weeks in some cases.
I am talking specifically about Deepseek.
If it doesn't know what year it is and when you challenge it, it doubles down and confirms the year is still 2024, then ANYTHING else it provides you is suspect. If you don't understand and agree with that you are a fucking idiot.
You didn't create Deepseek so stop squirming like I am trying to fuck you.
 
Then what are you expecting that AI to do for you?

I'm not even leaving it to only DeepSeek. But overall, any of them, what do you expect it to do or help you to do.

I use a few of them almost daily.
I've just never asked of it anything I'd already know as in what year it is. Seriously.

It's a tool to to simplify tasks for me that would take hours, days or weeks in some cases.
I asked it for information on Donald Trumps policies and it told me that Biden is the president.
When I said wait..... "What year is it"?
It said it is "2024".

Now @Mixd How the fuck and I supposed to get information on the year 2025 if the damn thing thinks is is 2024.

@HNIC @Madrox why did the 2 of you like his comment? Are you comprehending? Incorrect answers can never be tolerated.
Fuck a AI. If a person thinks it is 2024, you should have no confidence in anything they say! WTF!!!!!!
 
Question: Who is the mayor of Oakland California?

Deepseek:
As of 2025, the mayor of Oakland, California is Sheng Thao. She was elected in November 2022 and took office in January 2023.

Would you like more details about her policies or recent developments in Oakland?

GPT:
As of April 4, 2025, the mayor of Oakland, California, is Nikki Fortunato Bas. She assumed the role of acting mayor on December 17, 2024, following the recall of former mayor Sheng Thao.

Neither of these are true.
Both offer sources and volunteer additional information without any shame. Perfect for the "I did the research" era of the internet.
 
I am talking specifically about Deepseek.
If it doesn't know what year it is and when you challenge it, it doubles down and confirms the year is still 2024, then ANYTHING else it provides you is suspect. If you don't understand and agree with that you are a fucking idiot.
You didn't create Deepseek so stop squirming like I am trying to fuck you.
You right, you win.
Deepseek is trash cause it isn't up to date on 2025.

I'm not even on your level. No squirming here.
 
You right, you win.
Deepseek is trash cause it isn't up to date on 2025.

I'm not even on your level. No squirming here.
Good.
Deepseek is trash. Compared to ChatGPT and Perplexity it is trash.
 
Question: Who is the mayor of Oakland California?

Deepseek:
As of 2025, the mayor of Oakland, California is Sheng Thao. She was elected in November 2022 and took office in January 2023.

Would you like more details about her policies or recent developments in Oakland?

GPT:
As of April 4, 2025, the mayor of Oakland, California, is Nikki Fortunato Bas. She assumed the role of acting mayor on December 17, 2024, following the recall of former mayor Sheng Thao.

Neither of these are true.
Both offer sources and volunteer additional information without any shame. Perfect for the "I did the research" era of the internet.
See this is crazy if you never new the answer and asked AI to double check they would still give you a wrong answer.
 
Good.
Deepseek is trash. Compared to ChatGPT and Perplexity it is trash.
To be clear. They each have their own strengths. I barely use Deepseek. Earlier in the thread I thought they should be praised for making the software open source along with utilizing many more nodes with a lot less compute power than the others in its class.

Almost weekly things are getting better and better constantly changing. Example, these past few weeks many are saying Gemini is blowing away others. I mostly use Claude, and Grok. BUT, they are used for what I do as to what I found was the best. Haven't even touched Gemini yet, just was opened up to me last week.

So far since this thread started, I wrote a relationship self help book using Claude, it's 250+ pages, written under a pen name and completely humanized.
Then did 3 erotic novels on Grok cause it writes with graphic erotic sex. About 50-60k words each.

All under pen names.

Then did 10 adult coloring books, using Stable Diffusion but all the copy was with Claude.

Almost all the books are published on Amazon and the novels are also with major retailers and online libraries. Still editing two novels to humanize them and proof read

As I said earlier all have their strengths and weaknesses. I'm no where on a level that coders like my son who is a student game developer or others utilizing them to write complex software, data analysis or whatever else it's abilities.

took me a while to even know what and how much I can utilize these chat AI programs. Now I use them daily.

I used AI to research what were the top selling books, who my core audience was, how to market to them and what tools to publish them.

To me that's residual money, and royalties cause I own that.
 
To be clear. They each have their own strengths. I barely use Deepseek. Earlier in the thread I thought they should be praised for making the software open source along with utilizing many more nodes with a lot less compute power than the others in its class.

Almost weekly things are getting better and better constantly changing. Example, these past few weeks many are saying Gemini is blowing away others. I mostly use Claude, and Grok. BUT, they are used for what I do as to what I found was the best. Haven't even touched Gemini yet, just was opened up to me last week.

So far since this thread started, I wrote a relationship self help book using Claude, it's 250+ pages, written under a pen name and completely humanized.
Then did 3 erotic novels on Grok cause it writes with graphic erotic sex. About 50-60k words each.

All under pen names.

Then did 10 adult coloring books, using Stable Diffusion but all the copy was with Claude.

Almost all the books are published on Amazon and the novels are also with major retailers and online libraries. Still editing two novels to humanize them and proof read

As I said earlier all have their strengths and weaknesses. I'm no where on a level that coders like my son who is a student game developer or others utilizing them to write complex software, data analysis or whatever else it's abilities.

took me a while to even know what and how much I can utilize these chat AI programs. Now I use them daily.

I used AI to research what were the top selling books, who my core audience was, how to market to them and what tools to publish them.

To me that's residual money, and royalties cause I own that.
Oh I see. Fair enough. Very interesting.
So your work doesn’t rely on facts as it is creative and subjective. Fair enough.
That works for you.

My work relies on facts. Data needs to be accurate and clear and fast.

All of the AI I have used have been incorrect and I have to prompt it to double check its work.

Deepseek is the worst offender and when it is wrong it doubles down like a bitch and won’t correct itself.
 
Oh I see. Fair enough. Very interesting.
So your work doesn’t rely on facts as it is creative and subjective. Fair enough.
That works for you.

My work relies on facts. Data needs to be accurate and clear and fast.

All of the AI I have used have been incorrect and I have to prompt it to double check its work.

Deepseek is the worst offender and when it is wrong it doubles down like a bitch and won’t correct itself.
You'll find that a few of them don't or can't even get on the web.

Certain browsers like I use Brave on my cell and it will use AI to give you a brief quick search on what you're looking for. Gemini is doing the same for Google.

But like I said they all have their strengths and weaknesses.

Give you an example:
When I finished these novels I had them write, because they write chapter by chapter, it sometimes tends to lose the continuity of the story. So I put it all into a Word doc and uploaded it. Many of these chat tools has a limit of characters it can read when uploaded in chat. They are limited to somewhere between 140k-160k characters in an uploaded document. These books can be 300k-350k characters for a 50k word manuscript. So asked one of them, cant remember which, but it told me upload the doc into Google Docs or Dropbox and it will grab it and read it. My purpose was for it to proofread the entire novel for continuity, errors or anything. Which it does in 1-2mins. Imagine how long a person would take and the cost to do that such as an editor.

It read the whole thing. Told me which chapters were with redundant phrases or words, rewrote whatever it suggested was off (had to ask it to rewrite some sections) and it spit me back out chapters.

Point is I found some cannot even go on the web to search info or do certain tasks. They are limited to holding the data to when the version of that AI chat was released. I believe they probably dont want to corrupt the AI with a lot of wrong data on the web and it can learn to be better suited with real user inputs. All this is still growing pains.

They will iron things out.

But I just asked DeepSeek if it can search the web. It said it can if you enable the Web Search button, I did and asked it what year it was and it told me 2025. Turned it off and still spit back the year 2025.

I'd ask a multitude of chat apps the same question you're trying to get to and see which output best suits what you're looking for, then work with that. It's how I found that Grok allowed sexual outputs for the novels. None of the others I tried could.
 
Back
Top