Hello, everyone,
I am a ChatGPT Plus user and have been for quite a while. I love the option to select different models (I hope GPT-5 still gives some flexibility on this), but, as a subscriber to Claude and Gemini, I am forced to use them when dealing with big documents or having a long conversation.
Claude’s context window is 200k tokens and Gemini’s is 1M tokens. In comparison, ChatGPT has 8k on the free version, which I can understand as these things are expensive and you want people to subscribe if they’re using it for long tasks. However, a shockingly low 32k context window on Plus (and Teams) is disappointing. Even Pro (and Enterprise) is just 128k.
In my opinion, that makes no sense in 2025, especially when the competition is clearly ahead on this issue and OpenAI itself already has models that support longer context windows (up to 1M with 4.1 in the API).
32k is barely usable for long documents/chats, and I feel 200k should be the bare minimum for Plus (even though I’d argue it should be between 500k and 1M – Gemini and Claude don’t lock their bigger context windows to their highest-priced plan).
The other downside of such a low limit is that the model’s output tokens are capped, so some answers that should be really long and detailed are cut short because of the hard limit. If I am not mistaken, Gemini 2.5 Pro has 64k token output. I estimate ChatGPT is around 5–6k.
I urge OpenAI to reflect and improve on this. I feel it is far more important than allowing image generation on WhatsApp for free.
Thank you.
6 Likes
I upvote this agessively. It must be 128k at least.
2 Likes
Thank you. It’s ridiculous to be 32k for paying users and forces me to use other platforms. Makes even less sense when the models can actually handle bigger context windows and bigger output tokens.
2 Likes
Yes! Now I need to pay for Gemini, which i overall dont like, but the context is to good to avoid.
I totally agree and it seems that the information on actual context window in ChatGPT is really hard to obtain !
1 Like
Actually, I have some doubts about the accuracy, but here is an answer from OpenAI support on this matter..
I’m an AI support agent for OpenAI and I’m here to clarify things about context windows for ChatGPT Plus and API models.
For the ChatGPT Plus subscription (browser/web app), the latest GPT-4o and earlier GPT-4 models use a very large context window—currently up to 200,000 tokens for the o3 and o4-mini models available in both ChatGPT and API. This is much larger than 32k tokens.
For API usage, the o3 and o4-mini models also support up to 200k context window and can generate up to 100k output tokens per request.
Important: The much smaller (32k) context window was typical of earlier GPT-4 models (not o3/o4), but this is no longer a global limit. Some models, like GPT-4o and o3, support 200k tokens now. There are other models (like o1, gpt-4.5-preview, gpt-4o-mini, etc.) with varying context sizes, so it pays to double-check which model you need for your specific workflow.
1 Like
I would also like to see GPT go up to 1mill tokens. If it does I might just subscribe again.
1 Like
This is not true at all. For API, sure. For the ChatGPT app, 4o, 4.5, o3, o4-mini don’t have 200k tokens context window and surely 4.1 doesn’t have 1M. Not only that, their responses are limited, don’t get close to maximum output tokens. I know it because I tried the same prompt on API and the app and the responses were vastly different.
I also tried documents and in the app it struggles when they get big.
Thanks for the input…
I had the feeling that this was the case.
I think I’ll try to check having access to the API for high-token tasks.
Since your talking about 4.1, do you know if it is better than 4o in assisting to synthetize and analyze long documents, even without going to the 1M token limit ?
Just an update !
Got an answer from a human at OpenAI which clarified things a bit.
It seems most ChatGPT Plus Users have a 128k context window so far.
Difference with the API likely comes out from differences with the instruction (ChatGPT have custom instruction to act as a itself, consider memory, contextual memory from other chats and so on.)
Also, regarding differences in managing big documents, it’s my understanding that if you use the upload document using the upload file interface from ChatGPT, the whole document is not uploaded into the context window I think ; its like the file search API function, which chunk your document in possibly 800 tokens parts and then section linked semantically to the input are extracted and fed to the LLM for an answer… Thus the model do not see the whole document unles you copy the whole text directly in the context window (or canvas).
We understand that you came across the 32k token limit on our website, and we truly appreciate your feedback regarding our product information. Your comments have been noted and will be reviewed internally.
We’d also like to clarify the earlier mention of a 200k token limit — that specific limit applies only to our Enterprise and Education offerings. You can find more details here.
For ChatGPT Plus users, models such as GPT-4o and GPT-4.1 mini support a 128k token limit, which is significantly higher than the previously mentioned 32k. For additional information on token limits and model capabilities, please refer to this page.
Thank you again for bringing this to our attention. If you have any further questions, we’re happy to help.
I am very skeptical of that, especially because their own plan advertisement says context window is 32K. But, regardless, the output tokens are severely limited, unfortunately.
Yes, in my experience, it is. 4.1 is more “professional”, straight to the point, and better at following instructions than 4o.
However, 4o seems to have a “better” personality, more friendly, which is why I like to use it in the ChatGPT app.
1 Like
Yeah I think the limit on the output is 4096 regardless of the plan (except 4.1, but probably just in API). There really is some confusion between different websites … and their human and AI generated software. If I really spoke to a real human, I expect they’ll update their website soon lol. This confusion is more frustrating than just knowing straightforwardly that it is 32k or 128k lol.
According to my internal indirect tests, it’s around 32k for plus. Pity
