Skip to main content

Txt file limit should be increased to more than 32k tokens


Can i get increased limit for file size when i upload text file which has huge amount of text i get File too large (> 32k tokens): all_src.txt

this is very useful because i upload whole context of application to gemini 2.5 pro and then it’s easy to develop any thing becuase chat has full context of my application

Thanks!

Status: Completed22 comments

Log in to comment and vote

Comments22

  • Mark Florkowski

    TeamPinned

    We are working on a more expensive tier with support for larger pastes and extended context. It was untenable to support large contexts on the $8 price, unfortunately.

    • Apricot Astronaut

      I will upgrade immediately for this plan. Repomix → t3chat + gemini 2.5 pro is the dream. Happy to pay 20$/month.

      Unsure about costs for several million tokens of context.

    • Cyan Snake

      Seconding Apricot’s point. I would happily pay more to be able to paste in more context for the LLMs. The UX for t3 chat is excellent and I’ve enjoyed using it. but this constraint makes it very annoying when I’m working with larger projects.

    • Olive Penguin

      BYOK is for it, no? If you give an estimate for token usage, everyone will be very happy and its not going to cost you an arm serving large context

      • Mark Florkowski

        Team

        We do plan to lift limits for BYOK, just have not gotten to implementing this yet!

        • Silver Milkshake

          are you planning to remove the max conversation length limit too? the other day i had a long BYOK claude conversation and eventually got an error message that i had reached the maximum conversation length

          • Mark Florkowski

            Team

            Yes

        • Bronze Skateboard

          +1 this feature would be hugely helpful. I’m using o3-pro with BYOK because I have lots of OpenAI credits. This was my main reason for buying T3 chat premium. Sucks if I can’t make use of the long context on that model.

          • Mark Florkowski

            Team

            o3 pro has no limitations on context in t3 chat because it is BYOK only

  • Beige Cascade

    When implementing this, consider adding a feature where users can choose between having 1500 messages per month with 32k token window or 750 messages per month with 64k token window. Or something like described here:

    https://feedback.t3.chat/p/expanded-token-windows-with-proportional-message-deduction

  • Teal Lightbulb

    Any updates on this problem?

  • Beige Plasma

    I’ll pay more this as well. I just started using T3 Chat and love it, but this is the biggest blocker for me at the moment.

  • Chocolate Sand

    Pleaseee a higher plan for power users!!

    • Mark Florkowski

      Team

      In progress!

  • Coral Cucumber

    This might be doable if you change the usage to be pure per-token rather than message based, so it just eats more usage rather than not being an option at all. Plus, long chats take a lot more than 32k tokens anyway, plus inputs are cheaper, plus caching is usually 10x cheaper too. You don’t necessarily need to charge more, just make usage consumption smarter. It doesn’t seem a good idea to prevent one off large messages for everyone simply because it opens the door for profit loss. Super power users abusing the larger token limit is mainly an issue because of the message credit system allowing theoretically much higher usage, correct?

  • Aqua Iguana

    Yes, I will pay more for this feature, honestly

  • Coffee Syrup

    It would be great if the limit was split over an account’s usage period rather than a per limit. I will often get summaries of meeting transcripts that are 3+ hours long and with time stamps that can easily be over 32k tokens or even 64k tokens. I’m not against a more expensive plan, but only a couple dollars is the most I’d pay for larger file uploads.

    • Mark Florkowski

      Team

      Interesting idea, can look into it

  • Gold Beaver

    Now that we have Premier. Is this planned soon?

    • Mark Florkowski

      Team

      possibly