Why Custom GPTs are better than plugins

In my experience they are equally unusable due to compounding reliability issues.<p>It’s a tech demo, not a platform, a data acquisition operation and alpha testing rather than anything seriously useful.<p>Interfaces are unstable. Inference is intentionally non deterministic and control is crippled (e.g by choosing not to expose seed, denoising or image to image on dalle), to avoid PR backlash whole simultaneously hiding the true capabilities of the platform.<p>something as simple as using VITS to analyze an image fails 30% of the time because GPT5 decides it doesn’t have vision, wants to use pytesseract instead or writes hallucinatory pytorch code for a non existent vits library instead of just using inference.<p>One may create prompts that temporarily don’t fail at a high rate but constant silent finetuning, system prompt changes and unstable models &#x2F; APIs make the whole thing a tech demo designed to get users to volunteer future training data

Custom GPTs <i>are</i> plugins, just more streamlined. It&#x27;s still just a carefully written system prompt + some basic middleware scanning the output and occasionally taking over.<p>The UX may be different, sure, but there&#x27;s no <i>technical</i> difference and no <i>technical</i> innovation here. The main value of both plugins and custom &quot;GPTs&quot;[0] is that they&#x27;re <i>first party</i>. You can build or buy better implementations, but it won&#x27;t be <i>the</i> &quot;GPTs&quot;.<p>--<p>[0] - Kudos for whoever at OpenAI that approved calling those &quot;GPTs&quot;, for selecting a term that maximizes confusion not just about their offering, but screws with people&#x27;s comprehension of LLMs in general.

I love custom GPTs.<p>Superpower 1: Uploading Binaries and execute them i.e. ImageMagick <a href="https:&#x2F;&#x2F;chat.openai.com&#x2F;g&#x2F;g-j2c2iPuXI-franz-enzenhofer-chat-with-imagemagick" rel="nofollow">https:&#x2F;&#x2F;chat.openai.com&#x2F;g&#x2F;g-j2c2iPuXI-franz-enzenhofer-chat-...</a><p>Superpower 2: Treating any HTML page as API i.e.: Searching Google from ChatGPT <a href="https:&#x2F;&#x2F;chat.openai.com&#x2F;g&#x2F;g-jQApHmfQD-franz-enzenhofer-search-g-o-o-g-l-e-dot-com" rel="nofollow">https:&#x2F;&#x2F;chat.openai.com&#x2F;g&#x2F;g-jQApHmfQD-franz-enzenhofer-searc...</a><p>Superpower 3: Just automating annoying stuff i.e.: Was it a Google Update? <a href="https:&#x2F;&#x2F;chat.openai.com&#x2F;g&#x2F;g-1ceZagR5h-franz-enzenhofer-was-it-a-g-search-update" rel="nofollow">https:&#x2F;&#x2F;chat.openai.com&#x2F;g&#x2F;g-1ceZagR5h-franz-enzenhofer-was-i...</a><p>Or just a super well crafted prompt I use again and again <a href="https:&#x2F;&#x2F;chat.openai.com&#x2F;g&#x2F;g-WX2dWnIji-franz-enzenhofer-fast-data-visualization" rel="nofollow">https:&#x2F;&#x2F;chat.openai.com&#x2F;g&#x2F;g-WX2dWnIji-franz-enzenhofer-fast-...</a>

I am very surprised that many here don&#x27;t understand the great power and value of custom GPTs: They give you access to online APIs combined with pre-configured custom prompts and a compute sandbox!<p>You can query databases, trigger events and handle the results smartly.<p>Now that OpenAi added the @ sign to talk to your preselected custom GPTs you can just use different APIs like slack colleges:<p>@downloader get the data from test.tsv @sql create table according to tsv header @sql insert data @admin open mysql port<p>One thing to keep in mind is that even though custom gpts have access to a local sandbox file system, passing data around almost always involves GPT handling the data which becomes forbidding for any large token stream.<p>One critique that I share is the stupid branding &quot;costume GPTs&quot; and lack of discoverability: If you search the GPT store for wolfr you do not get wolfram alpha as completion! It only appears when you type wolfra<p>Also it can&#x27;t display any images other than dalle fantasies or pyplots which is a slightly annoying limitation, but familiar to users of other shells like bash.

Reading this article brought me no closer to understanding why people use these. From day one you could give any LLM any context you wanted, that&#x27;s the whole point after all.<p>The actual next stage of LLM development will be giving the user the ability to select&#x2F;deselect what training data to include otherwise there is a limit of at most a few pages of context you can provide.<p>Custom GPTs are trying to pretend thats what they are doing but it&#x27;s not going to work.<p>Like you can&#x27;t upload a novel and say, &quot;speak to me as if you are this character&quot; because the LLM can&#x27;t ignore it&#x27;s training data entirely and the context you give it gets drowned out quickly.

OpenAI has made very little effort on discoverability and filtering.<p>The opaqueness of Custom GPTs and the low effort in creating them compounds the problem.<p>1. Allow me to filter by &quot;feature&quot;. I want to explore only GPTs with custom APIs or uploaded knowledge. At least let me filter by &quot;length of custom instructions &gt; x&quot; so I can avoid 10,000 lazy submissions<p>2. Allow viewing of the custom configs by default. If an author chooses to disable this, then fine but &quot;sharing by default&quot; is a powerful mechanism to improve the ecosystem

I always feel like there is some trick to these I am missing out of, are there any good guides? Any time I look for some its just typical low effort blog&#x2F;youtube spam trying to get in on the AI&#x2F;GPT key words.<p>I have tried to work on one where I uploaded various documentation and spec sheets, wrote detailed instructions on how to search through it. Then described how it should handle different prompt situations (errors, types of questions, quotes from the documentation). It is able to search through the provided knowledge and provide quotes and responses with it, but it at no point gives a coherent response, so it basically always functions like a more intelligent search feature. Putting that it should re-prompt itself with the knowledge extracted and rationalize&#x2F;elaborated on it doesn&#x27;t seem to do much either, though it did provide some improvement.

Not a great comparison, custom GPTs use plugins or function calling similar to how they use code interpreter. You can&#x27;t really compare them because plugins or functions are a tool that GPTs use in addition to other features. The advantage a custom GPT has is it is easy to set up but it comes with big disadvantages such as having to use their RAG system which is very opaque and only being able to use one system message. Building with the assistant API can be far superior but requires a lot more effort and skill in building your own APIs.

I hated the custom gpt experience.<p>1) the chat transcript is lost when you close&#x2F;reopen the workspace. All that nuanced training conversation, gone.<p>2) the instructions for the gpt are just a summary of the training conversation. And those “instructions” were just too generalized- none of the nuance that was discussed.<p>I made a therapy gpt, but it simply wasn’t very useful after all of my instructions.

I mainly need one reason: Microsoft doesn&#x27;t hoard all your data

Think GPTs can scale way better than people realize. Things where you don&#x27;t always get good developer support or ecosystem (eg: GIS) and things that are ad hoc, GPTs w Code Interpreter can enable users to get value and get their work done without relying on a developer. The UX is well thought out and built with non techies in mind.

I feel the key question is not that if they’re better, but if people are using them more and more often than plugins.<p>And in particular, if they’re using the ones that aren’t just a custom system prompt. Because I really doubt there’s any big business in commercializing system prompts.<p>My hunch right now is that GPTs have made it clear that OpenAI should let user save multiple system prompts, but that there’s no real defensible business in distributing GPTs, as a chat interface is not that good for most purposes.

How do you tell the GTP to use the different category of files you upload in the knowledge base?<p>For example:<p>file A, File B : those are &quot;data of users&quot;, use them to do &quot;Y&quot;<p>file C, File D : those are &quot;data of buildings&quot;, do &quot;X&quot;