Context Window Size for PLUS users

ChatGPT Feature requests
,
1

Hey everyone,

I’ve been using ChatGPT Plus for a while now, and I have to say, the 8,000 token limit on ChatGPT-4o feels incredibly restrictive, even for basic tasks. It’s challenging even when you simply want to summarize something. It severely limits the model’s capabilities, reducing it to just the basics, like writing something simple such as an email. It almost feels as limiting as GPT-3 used to be.

I understand that API users have access to the 128k token limit, but it would be fantastic if Plus users could at least get 32k tokens. Ideally, this could be implemented in the ChatGPT-4o version, but even in the Mini version—since it’s cheaper for OpenAI to operate—it would be great if OpenAI took Plus users more seriously.

With alternatives like LLAMA 3.2 (which is free to use at Meta.ai) and LLAMA 3.3 (which is about to be released and already outperforms ChatGPT-4o in many cases), it’s hard not to feel the need for ChatGPT to keep up. While LLAMA doesn’t offer the full 128k token limit on the platform (though it does if run locally, which not everyone has the equipment for), it at least provides users with the same 8,000 token limit, without requiring a paid Plus subscription.

It would be fantastic to see at least 32k tokens available in the Mini version or a similar option. This would make a huge difference for paying users. Does anyone else feel the same?

2

The fact that they HIDE the actual context window and don’t clearly state it anywhere obvious is very shady and deceptive. I will never pay OpenAI money again. Not even the API. Fool me once… I’m not going to pay money to a company that shows it does not care about being honest and transparent with its customers. I totally get that you want me to pay for it and not use it unless I’m on the API and then you want me to use it just enough to have issues and need to keep using it. I can run models locally with over 100k context windows… offering a paid product with such strict limit on context window is a slap in the face. Antropic and Google both have a much better deal for their monthly plan users, giving the full stated context windows for the models….
How can anyone take you all seriously when even your presentation slides for GPT5 have obvious errors in them that none of your workers apparently noticed…. because I’m sure GPT5 made it and no one fact checked it!