When we talk about generative artificial intelligence, models and chatbots like ChatGPT, Bing, Claude, Bard and others, we usually focus on some basic and technical aspects about its specifications and operation, such as the maximum number of interactions per conversation, the amount of data used in its training, its speed, the means of input it allows… that is, those that are quickly measurable and that, of course, have a significant impact very directly in the interactions with them.

However, it exists other lesser known but equally important aspects, in some cases more, and popup is one of them. But in order to understand why the leap I mention in the headline of this news is something truly tremendous, it is necessary to first understand some technical terms. But don’t worry, they’re basic and interesting, and we’ll go over them in a moment.

The first basic concept is the token. If you’ve ever used a generative model, it’s quite likely that you’ve seen its output measured in tokens, not words, pixels, notes, or those metrics generally used to measure content type. which is displayed. Tokens are basic units of information that are used to represent and process data in generative AI models.

Tokens can be words, characters, symbols or pixels, depending on the type of data used. In other words, models learn using words, notes, and bits, but divide them into tokens. Conversely, they always generate their output in tokens, even though the result they deliver to us is logically structured to be readable/interpretable for us. In the case of generative text models, it is usually assumed (although highly variable) that ten tokens are between seven and eight words.

from his side, popup refers to the maximum amount of information, measured in tokens, that the model can process for a work session, as a complete conversation if we are talking about chatbots, or data that can be accepted as an input to create an output based on it. However, it is important not to confuse the data in the context window, which is specific to each work session, with the data used to train and validate the model, which is information that is usually provided when talking about it. and whose volume is gigantic.

So the maximum size of the popup window depends on the data we can give the model to generate output based on it. And of course another important metric is the time it takes the model to process this data. And it is at both points that Claude, the generative model of Antrophic and which you can try out in the chatbot mode in Poe, decided to hit the table, stand high above your competitors.

Specifically, as we can read in a post on his official blog, Antrophic raised Claude’s popup from 9,000 to 100,000 tokens, which the company estimates at roughly 75,000 words. To give some context (pun intended), the standard GPT-4 model is capable of handling 8,000 tokens, while the extended version can handle 32,000 tokens. ChatGPT has a limit of around 4000 tokens.

This increase enables e.g. Claude is able to devour an entire novel in one working session, which would allow, for example, to use it to check a complete design. Or also use it to view a large selection of reports to generate a new one based on their information. And for closer use, that the duration of the conversations can be extended much more before the model becomes confused because of all the data contained in it.