OpenAI’s New Program Explained

Good Morning AI Runners 🏃‍♂️

Here's what we've got for you today:

• AWS and Hugging Face: Make AI Open

• OpenAI’s “Foundry” Explained

OpenAI’s “Foundry” Explained

OpenAI has developed a new program called Foundry, which allows customers to run the company's newer machine learning models, like GPT-3.5, on dedicated compute capacity.

Foundry offers a "static allocation" of compute capacity to a single customer, with service-level commitments for instance uptime and on-calendar engineering support. Rentals are based on dedicated compute units with three-month or one-year commitments. However, the instances won't be cheap, with running a lightweight version of GPT-3.5 costing $78,000 for a three-month commitment or $264,000 over a one-year commitment. It seems pretty obvious that OpenAI is under increasing pressure to turn a profit after its multibillion-dollar investment from Microsoft.

Plus, there's some new mysterious model with a 32k context window that could be GPT-4, or a stepping stone to it.

Make AI Open ( for real this time )

AWS is collaborating with Hugging Face to make AI work easier on its cloud. If you didn’t know, Hugging Face is a software development hub where AI developers share open-source code and models. AWS will make Hugging Face's language generation tools, including a ChatGPT rival (Bloom), available to cloud customers for their own applications.The two companies are working closely on their shared products, but the deal isn’t exclusive! 

Pic of the day:

That's it from RunTheAI for today.

THANK YOU FOR READING AND SEE YOU TOMORROW, SUBSCRIBE TO STAY UPDATED!

P.S. if you made it this far, hit “reply” and tell me what you think of today's newsletter...what’d you love? What was boring?