← sources

OpenAI Is Giving Up on Its Best Feature.

source: reddit (reddit:OpenAI)published 2026-11-02 12:58ingested 2026-11-03 02:39

retained text

OpenAI is going to start imposing limits on conversations in December. This was arguably the best feature that set OpenAI apart from every other AI company, and now they're giving it up. They risk permanently losing a significant portion of their user base. Let me explain why.

I've been using ChatGPT since 2022. When ChatGPT first launched, it was introduced as an AI assistant, but its earliest users were mostly developers. The primary use case was writing code. OpenAI continued down that path until Claude emerged as a serious competitor.

After Claude came along, the industry realized something: the people willing to pay for AI were largely developers and professionals who actually made money using these tools.

Ask even the biggest OpenAI fan, and they'll admit that OpenAI has never really managed to surpass Anthropic when it comes to coding and software development.

Seeing its competitor gaining ground in this area, Sam Altman launched Codex and Tibo, trying to challenge Anthropic where it was strongest. Initially, OpenAI managed to attract users with generous usage limits and Tibo's likable personality.

But with the release of Opus, OpenAI started falling behind again.

What we're left with today is a fantastic, relatively affordable AI assistant in ChatGPT and a mid-tier coding tool in Codex.

It looks like OpenAI is giving up one of its biggest competitive advantages because of rising costs. But here's what I think will happen: usage limits will initially be generous, but over time, even regular users will start seeing limit warnings. We've seen this pattern ever since AI assistants first became mainstream.

Eventually, competitors and local AI models will become increasingly attractive, and users will gradually start leaving.

OpenAI will either have to abandon B2C entirely and shift its focus to B2B, or risk heading toward financial trouble.

AI infrastructure costs continue to grow as the user base expands. But I don't think the industry needs smarter AI models at this point. We've already reached a point where improvements in intelligence alone are no longer enough to satisfy users.

What the industry really needs is a way to run increasingly capable models on GPUs at a much lower cost.

Without that breakthrough, this business model simply won't be sustainable for any of these companies. Imagine if NVIDIA raised its prices tomorrow. How many people would still be able to afford using Claude or Codex?

I suspect we're going to start seeing serious problems on the infrastructure and hardware side of the industry as well.

document fetched 2026-11-03 00:28 · canonical https://reddit.com/r/OpenAI/comments/1xlwh31/openai_is_giving_up_on_its_best_feature