It's only been a year since their best and worst models were named o4 and 4o, respectively. They still don't seem to care much whether their model names make sense or are useful.
No I totally get that! I guess what my real complaint was, was that even after I went directly to their website because I was sick of the in-app vague Codex model descriptions, even once I was on OpenAI's "Compare Models" page, the natural language descriptions for each model were still incredibly vague and incredibly general.
Like I bet I could swap out the description for GPT-5.2-Codex and GPT-5.4 and you wouldn't even know a thing. I guess my expectation was just, if the model names are going to be awful, at least I'd hoped the actual descriptions on their website would be informative enough to fill the gap.
But in any case I think we should all stop supporting OpenAI anyway—I can't wait for my sub to lapse. Bad customer service is one thing but it rather pales in comparison to supporting the military.[1]
Especially while still maintaining an accelerationist position w.r.t. continued model development.
I wonder if it's partly because the descriptions are interdependent. As in, to keep them up to date, they'd need to update every description at each new release, since which is best for what will change.
OpenAI has been consistently making it hard to understand what each of their different models are for.
Frustratingly, their approach—as indicated by their recent UI decisions in the webapp to hide all settings for model reasoning effort and even the actual model name/type itself under: first, a collapsible element, and then, second, a second-up button for a pop-up menu, which then hides all of their other models (besides the latest) under a third collapsible element/button—to such complaints seems to almost always be to make things even less transparent by hiding even more things behind the UI.
Figure 1.1 What it takes now to (a) find what model OpenAI is making you use, (b) how much reasoning effort it is using, (c) change to an actually useful, older reasoning model like o3 that actually reasons, as opposed to 'adaptively' only doing so when it can spare compute.[1]
1. Issues/Complaints
Figure 1.2 What you see on the Codex app when you hover over each model to make its description appear.
The short one-liner descriptions in the Codex App that describe each model are really vague and unclear and do not give any information on what each model’s unique value proposition is or what they ‘do best’—which would help a user understand what they ‘do different’ from any of the other models.
All of the descriptions just seem like word salads or like someone fed an LLM a prompt to randomize and play around with a set of adjectives and nouns to come up with as many combinations of the same string as possible.
Par exemple :
1.1 ...This isn't even that hard
Maybe most annoyingly, it's not that something similar to this kind of model-by-model information break-down doesn't already exist. OpenAI has something kind of like this.
Figure 1.3 Proof that OpenAI already has the 'describe what each of your models do and what they do differently' capability and are simply choosing adversarially not to use it.
Although even here, OpenAI does not seem to be consistent or particularly... helpful. For example, if you want a more intuitive design setup for comparing models, on a completely different page, the 'Compare Models' feature gives you a nicer-looking way to assess their models, complete with cute lightning & lightbulb icons.
Figure 1.4 Comparison of 3 OpenAI models using the "Compare Models" part of their site.
1.2 How to Write Bad Descriptions
But this still isn't really what I'm looking for. Notice how absolutely inconsistent, vague, and useless the descriptions even here on their own website (above) are:
(Descriptions that contradict the serial/temporal order of their models)
(Descriptions for one model that really apply to all models)
(Naming schemes that may or may not indicate some special quality or difference)
2. Proposed Fix
What none of us want: OpenAI NOT to further reduce the range of model options/offerings to the end user or implement another model auto-router into Codex.
What I'd like: for the simple one-sentence model descriptions in Codex to actually be useful by describing what makes a specific model distinct from the others—beyond just the obvious fact that newer models are probably 'better' at coding.
e.g. A short 1-2 line sentence that answers the obvious question: why have they kept some of the other older models around?
In Short: Don't just describe the models—and don't even bother if you're going to do them this badly—but describe what makes each model different.[2]
(Request)
Does anyone remember when this literally used to be one single click for a drop-down menu?
Come on guys, this isn't really that hard.