5 models that aren't ChatGPT and what you can use them for
OpenAI isn't the only one having fun
Last updated: March 3, 2025
It’s now many months since I wrote about ChatGPT, and there’s even more going on in AI. OpenAI definitely took the world by storm, but open source models – and even closed source competitors like Midjourney – are catching up. Theoretical interest aside, this matters because these models can actually be useful to you at your job: generating marketing copy, cover images, doing research, writing your blog posts for you, you name it.
In the post about why ML models seem to be getting so much better, I wrote something I think is important:
Over the past couple of years, models have gone from private, code first, and inaccessible to widely available to the public.
If you look at models that have made it into the public discourse recently, like DALL-E or Stable Diffusion, they share a unique quality: whoever built the model (or their friends) also built an interface to the model. Using ChatGPT is as simple as typing a prompt into OpenAI’s website; generating a photo with DALL-E is too. It’s for everyone! And that is very weird.
In other words: models are getting better, but they’re also getting more accessible. On the cutting edge of research though, there’s less of a concern about how you – assuming you’re not someone who can code – can use the model. When a new model gets released, the interface is usually a bunch of code and some model weights, not a slick web application you can sign up for. So some of these models will be more difficult for the average person to use.
Anyway, this post will run through 5 non-OpenAI ML models, what you can use them for, and how you can get started.
Terms Mentioned
Companies Mentioned
Stable Diffusion
Who made it: StableDiffusion was developed by German researchers at LMU, in conjunction with RunwayML (a dope video editing tool). For the not-faint-of-heart, you can read the original research paper here.
What it does: it’s a text-to-image model, so it takes text and generates images. You can work iteratively with it, so once it generates an image, you can tell it to remove certain things, change others, etc.
StableDiffusion has parameters that you can mess around with to make it better match your use case. One is the “Guidance Scale,” which basically controls how tightly the model tries to match your prompt. Another one is “Steps,” which determines how many iterations the model does before it gives you the final image.
How to use it: since StableDiffusion is open source, there’s no official “website” that hosts it where you can sign up (unlike OpenAI). You can run it locally on your computer if you’re technically inclined, but the easiest way to use it is through one of the free online hosting providers like Stable Diffusion Online, Hotpot, etc.
LLaMa
Who made it: Facebook (I refuse to call them Meta) released LLaMA back in February of this year, and updated it in July.
What it does: LLaMA is a large language model, and generates text. You give it a prompt, and it writes for you, just like ChatGPT . What makes LLaMA interesting, besides it being really good and open source, is that it’s available in multiple differe...