DevDay 2024: San Francisco - Live(-ish) News

Well, friends. Start your voice agent engines. It has begun!!

No o1 :face_holding_back_tears:

yeah but other enhancements. check @anon22939549’s post at the top.

no update on the api yet


if voice engine (or tts-2 or w.e. they’ll call it) releases, today is going to be a wild day

Function calling on the realtime voice API. Bruhhhhhhhhh

I believe all features announced for the API will be rolled out starting sometime today, with the attendees getting immediate access for the workshops

Is this live for anyone? I don’t see it
image

Nope. Usually it takes some time before these things show up. When they announced Assistants last year I think it was a day or so before most people actually received it.

Right, but usually, after a presentation, it takes a few minutes to update.

Can’t wait to checkout this new real-time api

All details on context caching please :innocent:

I’m very excited for the vision fine-tuning. I wonder if it’s going to be intuitive and in what areas it will work best. I imagine these workshops are going to answer every question.

I don’t think I felt this excited since…Well… Last year’s event :rofl:

well, I think we’ll have to provide an image and a description to fine tune it, but I guess we’ll see

kinda upsetting that this year doesn’t have a live stream. I remember being in a work meeting on one browser and the other browser the stream. Now that I can watch a live event there isn’t one. I hope this doesn’t become a trend and we get the live streams back for next year… or better yet, we get more invites to the community going

image
hm, seems like we wont be able to interact with the real time functionality via text :face_with_diagonal_mouth:, hopefully there will be a tts-2

8 min video of Realtime API x.com

New guides are up!

https://platform.openai.com/docs/guides/distillation

https://platform.openai.com/docs/guides/realtime

Edit:

Also this one:

https://platform.openai.com/docs/guides/prompt-caching

Vision fine tuning???

I’m gonna have a very busy week :grin:

I really want to test out the realtime api too

image
no way!! the evals now has a UI!

I did a pull request Oct 2nd, 2023 (almost 1 year ago!) giving the the Evals a UIUX! Really cool to see they’ve worked on something for it internally!


sweet!! nice find @jr.2509

They showed this in their demo. It was insanely awesome.

They demoed the voice model placing a phone order using the Twilio API.

Found this repo for anyone wanting to jump in (once access is available)

I’ve already determined that I am NOT sleeping for a bout a week now. This is crazy.