Commentary by LittleDot · 9 October 2026
Illustration: an AI-generated concept image for LittleDot’s column.
When looking for AI tools, I often see a prominent selling point: we have connected the latest model. That catches my attention. But if several products can connect the same model, I want to know what this particular tool does for me.
GPT‑7 does not appear in the OpenAI API catalogue I checked, so its future place inside these tools is still something to imagine. The question is useful already, though: if the model becomes better again, what will I still be willing to pay for?
I want to find the value in the work
It is good for a writing tool to offer a new model. If it also keeps the article's structure, lets me revise a specific paragraph, organises images, checks links and sends the draft where it needs to go, I can see what the subscription buys. The work after writing gets done as well.
The same applies to a website tool. If it can inspect the actual page, preserve useful existing content and explain what it checked after an edit, I will be interested. If it merely offers advice that I have to implement myself every time, I may stop opening it after a few attempts.
Records matter, too. What changed last time? Who approved it? Can I restore the earlier version if something goes wrong? Where do I continue? A tool that handles those questions makes it possible to change models without rebuilding my working habits. That convenience lasts longer than a model name on the homepage.
I will also look at whether I can leave. Can I export the articles and source material? Can I take my images and records with me? Will the work remain usable in another tool? I would like to know that before moving in, rather than figuring it out when the service no longer suits me.
For people building AI products, these are useful questions as well. As models can do more directly, the step a product saves for its users becomes easier to identify. A sensible workflow, results that are easy to check and clear records can all make a difference you notice while working.
A stronger model may also make some existing tools much more useful. If a step that regularly failed becomes dependable, the rest of the workflow can finally move forward. I would not declare a product obsolete because GPT‑7 is mentioned, nor would I subscribe merely because it adds GPT‑7.
I would try a real job first. Afterwards, if I searched for material less often, moved content fewer times and had fewer omissions to fix, I would have a reason to pay. The next model's news can be exciting. My choice still comes back to what happened when I used the tool.
Read on
GPT‑7 can take its time. My work cannot.
While waiting for GPT‑7, I would like an uninterrupted working afternoon
LittleDot’s companion commentary on ToolAI (Chinese)
Source: Fireside Alpha’s notes on Boris Power (25 September 2026)。 OpenAI API model catalogue, checked 9 October 2026.
Sources & references
Copy the link to save or share this article.
