If you’re going to DevDay on the 6th November and whish to discus meetups, breakouts, anyting related to the day, you can do so here, feel free to contribute if you’ll be watching the keynote from the livestream on Youtube https://youtu.be/U9mJuUkhUzk
Totally hope to watch the livestream.
Any rumors on what the announcements might be?
Not heard anything yet, I’m going to be there, so I’ll post away if I find any interesting bits of info.
Interested to see what people think might be announced.
Let’s be overly optimistic and hope for a “GPT-5 with video capabilities” timeline.
I’m just hoping for “GPT-4 Vision” API access for image inputs
But wow, GPT-5 API announcement? Insane that would be.
Or like “Dalle 4” would have text to video
(PS it’s GPT-V)
You can tell its in the works - the recruitment of new red-teaming.
Whatever the next model is called, switching over to Roman numerals would be super confusing, especially now that “V” means “Vision”.
I don’t think so, sorry to dull the vibe, but I think there might be some mention of 4.5 but I don’t think 5 is going to be mentioned much, if at all. I hope I’m wrong! I’d love it to be AGI, but I think there is so much work still to be done with GPT-4 class models and the steep cliff in terms of hardware to run models on to be climbed.
Multimodal will be a big part of it, not sure about video yet, maybe some early stuff. Personally I’d like to see work on allowing large datasets to be reliably parsed being a priority.
Following the news I now fully expect a big announcement of the model with codename Gobi.
It appears this is the original source:
But using Google there are short write-ups available.
I am expecting (and hoping) that they will release the API for Vision on DevDay. And also wishing that they drop us devs a holiday gift by releasing the Text-to-Speech API (let’s call it Echo® ) before the year ends lol.
Releasing Dall-E 3 and Vision for API would be insane. Fingers crossed.
These AGI rumors are also very exciting. What a race.