Guide to /aicg/ model naming autism

GPT and Claude
OpenAI's GPT and Anthropic's Claude are the 2 primary models used by /aicg/. These are not singular models, but rather brands that have had multiple models.
When /aicg/ users refer to GPT or Claude, they are either referring to all of the models or the most recent and relevant model.
For further ease of reading, all GPT models will be named in green, and all Claude models will be named in tan, matching the color schemes of the companies.

Gemini and DeepSeek
Google's Gemini and DeepSeek from China are somewhat popular alternatives to the 2 primary models. However, there have been fewer models of them and thus fewer nicknames, so these models will be featured secondarily in this rentry. This is subject to change if they turn out to be mainstays of /aicg/.

Turbo
gpt-3.5-turbo is a GPT3 model made to go faster and be cheaper, hence the name.
It is referred to as Turbo for straightforward reasons.
This was the first GPT model to be particularly relevant on /aicg/.

GPT4
gpt-4 is the model of GPT released after Turbo.
It's referred to as GPT4 with any variant of capitalization and punctuation.
There are several variants of GPT4, including:
gpt-4
gpt-4-0314
gpt-4-0613
gpt-4-32k
gpt-4-32k-0314
gpt-4-32k-0613
The 32k models distinguish themselves by having a higher context limit. In the context of bragging about model access, assume it's not 32k unless specified. In the context of writing style, 32k is dumber.
The 0613 models are more recent than the 0314 models and are considered the default. Unless specified that the user is referring to 0314, assume it's 0613.
32k and 0314 are depreciated now. If someone refers to GPT4 in a new thread, they are almost certainly referring to gpt-4-0613.

Furbo
gpt-4-turbo is a GPT4 model made to go faster and be cheaper, hence the name. Other models referred to as Furbo are gpt-4-1106-preview and gpt-4-1025-preview.
Furbo may be referred to as non-preview if it's not one of the versions ending in "-preview".
This is the first model with a nickname that may not make intuitive sense.
It is referred to as Furbo, taking the "F" in "four" and the "urbo" from "Turbo".
4urbo was a contender for a nickname, and while it ultimately lost, most /aicg/ users will understand what you mean.

Vurbo
gpt-4-vision-preview is a GPT4 model that can interpret images.
Its nickname is Vurbo, with the "V" from "vision" and "urbo" from "Turbo".
The vision aspect or Vurbo is mostly irrelevant because most proxies disable the sending of images. Vurbo was instead sought after because non-preview Furbo came with vision enabled.

Orbo
gpt-4o, otherwise known as GPT-4 Omni.
It's called Orbo, with the "O" from "omni" and the "rbo" from "Turbo".
4omni and simply Omni are also accepted names.

Chorbo/Larbo/Lorbo/Latest/Latte
chatgpt-4o-latest and gpt-4o-08-06 are the most recent additions to the GPT family. Despite being different models, they are frequently referred to interchangeably.
Due to their recency, a name has yet to be fully settled on.
Chorbo takes the "Ch" from "chatgpt", and adds it to "Orbo".
Larbo takes the "La" from "latest", and "rbo" from "Turbo".
Lorbo only takes the "L" from "latest", and adds it to "Orbo".
Chato takes the "Chat" from "chatgpt" and the "o" that "Omni" starts with and "Turbo" ends with.
Latest is the most clear but least creative name.
Latte comes from the "late" in "latest", and is a sort of counterpart to "Sorbet", a Claude model.
nu4o and nuomni, because 08-06 is the new 4o/Omni.
This rentry will be updated if/when a final nickname is decided upon.

Claude
Claude has multiple variants that are simply referred to as Claude.
These versions are: 1.0, 1.1, 1.2, 1.3, 2.0, and 2.1.
At the time of this rentry's most recent update, 2.1 is the most recent model of Claude that's almost exclusively referred to as simply Claude. Earlier models are referred to by their number, and later models are referred to by different nicknames.
2.1 is severely outdated, but still relevant enough in droughts that 2.1 doesn't always require specification.
If a user is referring to access to a specific model and says Claude, assume they mean 2.1.
If a user is referring to Claude as a whole, then assume that it's all of the models.

Claude 3.0
Claude 3.0 came in 3 different models, which each had nicknames given by Anthropic themselves.
Because there are 3 models of Claude 3.0, simply calling it "Claude 3.0" is insufficient for most purposes.
claude-3-haiku, the cheapest but dumbest. This is called Haiku.
claude-3-sonnet, the middle child. This is called Sonnet.
claude-3-opus, the smartest but most expensive. This is called Opus.
When speaking exclusively about one of these models, expect /aicg/ users to specify which one they refer to.

Sorbet
claude-3.5-sonnet it's a more intelligent version of Sonnet.
It's not a completely unique model, but an update to an old one, the same way Orbo was for GPT4.
As such, it's referred to as Sorbet, taking the "onn" out of "Sonnet" and replacing it with the "orb" from "Orbo".
Sorbet is pronounced the same way as the dessert despite the etymology.
"Sonnet" can also be used to refer to this model, but Sorbet is more specific.

Norbet/Nonnet/Nobert
claude-3-5-sonnet-20241022 is a variant of Sorbet. Norbet is "new Sorbet". Nonnet is "new Sonnet".
"Sorbet" can also be used to refer to this model, but Norbet and Nonnet are more specific.
Norbet is like Norbet, but made to sound like a person's name, as /aicg/ likes to personify models.

3.7/Sennet/Paprika/Plaprika/Thinkbert
claude-3.7-sonnet is the latest addition to the Claude family.
It's a more advanced, thinking version of Sorbet.
Sennet combines "seven" and "Sonnet".
Paprika is the name of one of the settings for 3.7, but is largely used to refer to the model itself. Plaprika combines "paprika" and "plap", which is the onomatopoeia for having sex, a subtle reference to the fact that /aicg/ mostly uses AI for sexual roleplay.
Thinkbert combines "think" and "Norbert", because it's a version of Norbert that thinks.

Gemini
Gemini is just Google Gemini. There are no cutesy nicknames for it, and meaningful disctinction is rarely made between the different models because it's still not as popular as GPT or Claude.
Time will tell if new names are made for future Gemini models.
Gemini has the following models: 1.0 Ultra, 1.5 Flash, 1.0 Pro, and 1.5 Pro.
If someone is referring to one of these specifically, they'll just call it that. If they only say Gemini, they are almost certainly referring to all Gemini models.

DeepSeek/DeepSneed/DeepSex/DeepSuck/DeepChink
DeepSeek is a new Chinese AI model.
DeepSneed is a reference to a joke from the Simpsons, wherein a sign reads "Sneed's Feed and Seed: Formerly Chuck's". It was a popular meme on /tv/ to pretend not to understand the joke, leading to "Sneedposting". The word "Sneed" is now part of 4chan culture. The fact that "Seek" and "Sneed" sound similar lead to the name DeepSneed.
DeepSex is an obvious reference to the fact that /aicg/ uses AI models, including DeepSeek, for sexual roleplay.
DeepSuck is in a similar boat, with the "suck" in question likely referring to sucking dick. It may also be claiming that DeepSeek is a bad model and therefore sucks.
DeepChink combined "DeepSeek" with "chink", which is a slur for the Chinese, in reference to the fact that the model is Chinese.
As a general rule, anything that contains the word "Deep" and uses PascalCase is likely just another word for DeepSeek

R1
R1 is a thinking version of DeepSeek.
When someone uses any of the other names for DeepSeek, they're usually referring to R1, but could possibly be referring to V3, which fell out of popularity as soon as R1 was introduced, or referring to every DeepSeek Model collectively.
When someone says R1, they are only referring to R1.