ChatGPT Store Opportunities, GPTs, and Discussion

[uhhfeef]

GPT Store launched. Anyone here releasing their own GPTs?

Introducing the GPT Store

We’re launching the GPT Store to help you find useful and popular custom versions of ChatGPT.

openai.com

 

Dislike ads? Remove them and support the forum: Subscribe to Fastlane Insiders.

[MTF]

GPT Store launched. Anyone here releasing their own GPTs?

Introducing the GPT Store

We’re launching the GPT Store to help you find useful and popular custom versions of ChatGPT.

openai.com

View attachment 53457

In Q1 we will launch a GPT builder revenue program. As a first step, US builders will be paid based on user engagement with their GPTs. We'll provide details on the criteria for payments as we get closer.

That sucks. I hate it when everyone else but the US is excluded. Once it's open to everyone the opportunity will be gone for non-US people.

 

[alexXx9]

That sucks. I hate it when everyone else but the US is excluded. Once it's open to everyone the opportunity will be gone for non-US people.

yeah, we "outsiders" are almost always the ones who will eat the scraps of other, such as Audible on Amazon or Amazon Influencer program...tons of people making money passively and we European can only watch

 

[msufan]

The GPT Store (Explore GPTs) might be worth its own separate thread.

First of all, for anyone who has an existing entity, get on here! It's easy to imagine the Fastlane AI that helps you design your next venture (MJ) or the Google Ads AI that helps you become an ad expert (Andy) or anything else you already do.

Then there's selling the shovels: anyone who can make a GPT get to the top of the store will make millions -- either by getting their own GPT to the top or helping others to do so.

Then there's the whole market of helping people create their own GPT models in the first place. This will be HUGE as well.

This is equivalent to the Apple Store opening in my opinion. Time to act!!

 

Dislike ads? Remove them and support the forum: Subscribe to Fastlane Insiders.

[MJ DeMarco]

The GPT Store (Explore GPTs) might be worth its own separate thread.

First of all, for anyone who has an existing entity, get on here! It's easy to imagine the Fastlane AI that helps you design your next venture (MJ) or the Google Ads AI that helps you become an ad expert (Andy) or anything else you already do.

Then there's selling the shovels: anyone who can make a GPT get to the top of the store will make millions -- either by getting their own GPT to the top or helping others to do so.

Then there's the whole market of helping people create their own GPT models in the first place. This will be HUGE as well.

This is equivalent to the Apple Store opening in my opinion. Time to act!!

Moved to a new thread.

 

[MTF]

I might be late to the party but custom GPTs are blowing my mind.

Here's just one example, an RPG-style game with GPT. You can basically live your own adventure. No book needed, no gamemaster, no friends.

The GPT shows you a picture with a description and asks what you do. Then it responds based on your action with another image and text. Imagine once in the future it's taken from image and text to video...

 

[MJ DeMarco]

I might be late to the party but custom GPTs are blowing my mind.

Moved to the thread. But yea, the power of some of them is incredibly awesome.

 

Dislike ads? Remove them and support the forum: Subscribe to Fastlane Insiders.

[Saad Khan]

I haven't followed up with GPT updates since last September. It's crazy how far it progressed. I have a lot of catching up to do

 

[Xavier X]

The custom GPT Store seems a bit selfishly conceived by OpenAI, as only paid users can access them.
If only paid users can access them, most are useless, since the standard GPT-4 can do most things "custom GPTs" can.

Most creators simply can't add robust enough new data/knowledge base to solidly differentiate their custom GPTs.
So it'll just be the same guy (GPT-4) wearing a new hat each time. DoctorGPT, LawyerGPT, StartupGPT.

Someone correct me, if there's something I'm missing.

This is not to say there's no value in creating one, especially early in the game now. Down the line, I'm sure OpenAI would have to open it up to free users. For now, it's mainly just a conversion tool to encourage free users to get the paid version.

 

[Kavin]

That sucks. I hate it when everyone else but the US is excluded. Once it's open to everyone the opportunity will be gone for non-US people.

i mean cant you use a vpn and access it?

 

Dislike ads? Remove them and support the forum: Subscribe to Fastlane Insiders.

[MTF]

i mean cant you use a vpn and access it?

This is not about accessing their store but about being eligible for payments. They may require a proper US address and/or prove that you're a resident.

 

[Andy Black]

The custom GPT Store seems a bit selfishly conceived by OpenAI, as only paid users can access them.
If only paid users can access them, most are useless, since the standard GPT-4 can do most things "custom GPTs" can.

Most creators simply can't add robust enough new data/knowledge base to solidly differentiate their custom GPTs.
So it'll just be the same guy (GPT-4) wearing a new hat each time. DoctorGPT, LawyerGPT, StartupGPT.

Someone correct me, if there's something I'm missing.

This is not to say there's no value in creating one, especially early in the game now. Down the line, I'm sure OpenAI would have to open it up to free users. For now, it's mainly just a conversion tool to encourage free users to get the paid version.

What if we create them for our own use?

 

[AceVentures]

Get LM Studio on your PC. Here you can download huggingface OpenSource models and run local LLMs on your machine with a clean UI. This gets you 90% of the way with most LLM capability. There is vision capability but it's limited in LM studio at the moment.

There are also dozens of wrappers for different diffusion models.

There are many user interfaces for linking and piping different models together. I think node based programming interfaces are ideal for this type of work - and I use them accordingly. Touchdesigner is a super powerful software I use for interfacing with these models - check it out if you've never heard of it.

Play with NDI and OSC protocols for piping data between devices. For instance you can create a UI for the phone or ipad to interface with the model running on your PC using OSC to send/receive data. The model's memory runs on your GPU and you can use NDI which provides access to your GPU memory across applications.

It's fun to experiment with - and the pace of opensource innovation really levels the playing field. It means that any day, anybody can pick up a brand new model and develop a unique engine.

 

Dislike ads? Remove them and support the forum: Subscribe to Fastlane Insiders.

[Andy Black]

Get LM Studio on your PC. Here you can download huggingface OpenSource models and run local LLMs on your machine with a clean UI. This gets you 90% of the way with most LLM capability. There is vision capability but it's limited in LM studio at the moment.

There are also dozens of wrappers for different diffusion models.

There are many user interfaces for linking and piping different models together. I think node based programming interfaces are ideal for this type of work - and I use them accordingly. Touchdesigner is a super powerful software I use for interfacing with these models - check it out if you've never heard of it.

Play with NDI and OSC protocols for piping data between devices. For instance you can create a UI for the phone or ipad to interface with the model running on your PC using OSC to send/receive data. The model's memory runs on your GPU and you can use NDI which provides access to your GPU memory across applications.

It's fun to experiment with - and the pace of opensource innovation really levels the playing field. It means that any day, anybody can pick up a brand new model and develop a unique engine.

Ha. It's been a while since I read something that made me feel this hopelessly clueless.

 

[MJ DeMarco]

Ha. It's been a while since I read something that made me feel this hopelessly clueless.

I was gonna tell the forum about the new device I bought which uses a forced Matrix LLX model, utilizing the newest 45.1 GPUs which forces efficiency at the 200mh frequency at 5 ohms. Then you can run this device in alignment with Cam rod phases and 3.33 gearing for rapid acceleration up to 51 planks. And with a mega-class 3-tier docilometer, you can run it concurrently at 3.2 megajoules.

 

[socaldude]

I was gonna tell the forum about the new device I bought which uses a forced Matrix LLX model, utilizing the newest 45.1 GPUs which forces efficiency at the 200mh frequency at 5 ohms. Then you can run this device in alignment with Cam rod phases and 3.33 gearing for rapid acceleration up to 51 planks. And with a mega-class 3-tier docilometer, you can run it concurrently at 3.2 megajoules.

I was like dang, I've always suspected MJ was like an ultra high-intelligent undercover alien or something.

 

Dislike ads? Remove them and support the forum: Subscribe to Fastlane Insiders.

[AceVentures]

Ha. It's been a while since I read something that made me feel this hopelessly clueless.

LLM = Large Language Model
LM Studio = software that allows you to run LLMs locally on your own PC
huggingface = I think of it as github for machine learning models
GPU = Graphics card
UI = User Interface
Diffusion Model = Machine learning model used for generating images
Node-based programming interface = visual programming software in which you lay out your system into nodes and pipe them through different functions
NDI = a video connectivity technology which makes networking data more easy
OSC = Open Sound Control - a protocol for encoding and transporting arrays of data in real-time

Check out the attachment. It's a video I made yesterday for fun. It's a macro recording of my eye, which I do as a hobby. I've textured it with noise around the borders and I've added some displacement shapes to create more randomness. I pipe this video through an open-source tool called StreamDiffusion, which convolves the image and returns it back. It is surprisingly fast, it runs at ~13 frames per second. I use NDI to take that output from the model and bring it into my node-based programming environment. From there I can pipe the output forward and do whatever else I want with it. In this case I used Nvidia's upscaling AI model to increase the resolution, and then I created a layout as well as an option to export the data as a movie.

I'm not an expert
I'm not a computer scientist
I'm just a dude who is willing to click around long enough to create something fun

Earlier you wondered maybe we can use these AI Agents for our own uses. That's also one of my motivations.

You teach yourself random things all the time. You're always learning am I wrong? You learn about Google Ads, you learn copywriting, you learn the effectiveness of different ad strategies, you learn color theory maybe because you're a nerd and you wonder if it'll help you make ads with better conversion, you learn how to use Zapier so you can automate some of your email tasks, and on and on. Why couldn't you also learn this?

Just click around and see what you figure. On your own timeline.

 

[Andy Black]

Thanks for all those definitions @AceVentures.

Why couldn't you also learn this?

I'm sure I could, it's a distraction though and I have no pressing need.

I have wondered if a GPT created from all my forum threads could write in my voice, with my values, and my goals. (I don't wonder enough to do anything about it though!)

 

[AceVentures]

Thanks for all those definitions @AceVentures.



I'm sure I could, it's a distraction though and I have no pressing need.

I have wondered if a GPT created from all my forum threads could write in my voice, with my values, and my goals. (I don't wonder enough to do anything about it though!)

Same - no pressing need.

The diffusion stuff I do for fun. The local LLMs is so I can experiment with getting different results than ChatGPT, and also for being able to run it offline with fewer privacy concerns.

We learn whatever we're curious enough about.

 

Dislike ads? Remove them and support the forum: Subscribe to Fastlane Insiders.

[Eurojanek]

Your GPT buddy, smartphone killer is coming

View: https://youtu.be/Rqh6fhcAqpw?si=CMmOfAby6Wf-_U4k

 

[Xavier X]

What if we create them for our own use?

To create one for your own use, you first have to get the plus subscription.
To keep using it... you have to keep the plus subscription.