OpenAI’s Foundry will let customers buy dedicated compute to run its AI models

In the news
1

Hey everyone,

I’ve heard about this OpenAI’s Foundry will let customers buy dedicated compute to run its AI models

text-davinci-003 said I could get invited: "OpenAI Foundry is an invitation-only platform. To gain access, you must be invited by an existing member. " which I found curious since it wasn’t trained to respond to this I’d imagine. It had default values and zero temperature.

Oh, and just in case those invites exist, I’d love to have one in case someone has one sparing!

(also, any thoughts on how you guys would use it? I have a education company that was incorporated last year, I’m sure the team would love to check out the foundry on it’s beta state.)

3 Likes
2

Sounds interesting. Thanks for sharing!

Instances won’t be cheap. Running a lightweight version of GPT-3.5 will cost $78,000 for a three-month commitment or $264,000 over a one-year commitment. To put that into perspective, one of Nvidia’s recent-gen supercomputers, the DGX Station, runs $149,000 per unit.

Ouch! Maybe I’ll get there one day. :wink:

ETA:

Eagle-eyed Twitter and Reddit users spotted that one of the text-generating models listed in the instance pricing chart has a 32k max context window.

Wow, @raymonddavey … pricey, but imagine 32k context window!

1 Like
3

This is excellent news. I am a physician and looking at different LLM / GPT use in healthcare. I love this technology. I am the author of Chatgpt and healthcare.

4

This is fantastic info. Thank you! Not in my current budget, but would be nice to get there!

I wonder if one could do GPT 3.5 embeddings this way. That would be something quite valuable if it were the case.

5

Also, Davinci 003 makes up a ton of nonsense. Never ask it a serious question and expect a factual answer. It’ll agree to a game of pick-up basketball if you ask it nicely enough.