In the Generate Text action, you’ll find a new High Power v2 model, which is GPT-4o under the covers. It should be better in every way, and less expensive for processing large prompts.
If you are using Generate Text, please give this new model a try and report what you find. We’re currently evaluating this model for use with the other Glide AI actions as well.
We’ve also introduced Image to Text v2. Please give it a try and share your feedback. This is a huge leap forward from the previous version, in terms of vision and reasoning capabilities.
We had a customer attempting to process photos of handwritten forms, and it simply did not work with the previous Image to Text. This new version completed the tasks perfectly on the first try.
I was testing the “Complete chat” column and I can confirm that the new 4o model works just fine, is fast, and also costs less than 4. In this picture, you can see the costs of analysing a medium-long text through the API (around 4 pages in a document): I did it once with 4, and once (!) with 4o, and several times with 3.5-turbo during that day.
Next thing I will test soon is the capability to analyse a picture, which they claim 4o can do.
I’m new to ChatGPT subscription but just signed up recently to test 4o and use the new model. So if I use it on their site, I’m getting up to date information but the model through Glide AI integration set to High Power v2 is only showing up to Oct 2023. I know according to their docs, it says training data up to Oct 2023, but unsure how that’s different than the 4o I’m using through their desktop app or site.
I didnt see this thread on time and posted on another one…
But here is my questioning:
No double billing please… Let us use OPEN AI action with maximum capacity.(token windows etc)
From team plan maybe? else users only have access to Glide AI…
I just saw that:
Which would mean 96 updates (1,92$) to have one High power v2 request with max tokens (128000)…
I’m just curious, how many use cases you would have that require 128,000 tokens? That’s a lot of words and to be honest if you generate, say 124,000 tokens (taking out your input and system prompt), 2 dollars is not a crazy price.
128k tokens means 93k words. I did a search, let’s say a page has 300 words. That makes it 310 pages of content.
Whilst those models can have a maximum token window of 128k tokens, they have a limited output token. For gpt-4-0125-preview, that limit was 4096, so your question is valid. gpt-3.5 has this same limit stated on their website. On Microsoft’s website, gpt-4o is confirmed to have the same limit. Claude 3 as well.
The thread you linked to is GPT-4o, OpenAI released a mini version of it which is less expensive. I hope it will get integrated maybe in the next 2 weeks.
GPT-4o, the latest model from Glide, is now available" for a sec LMAO
