Paid for premium and still cuts me off on leo ai

Brave Browser Leo AI
1

paid for premium and still says i reached limits. Do i need to change something in settings or what? paid for premium not to be limited like free version kinda confussed. someone help

2

Hello @ChurchofYahweh - typically, if you’re running into limits, we’d suggest switching the model you’re currently using. However, in order to troubleshoot further, could you let us know the following?:

  • What specific OS you’re using?

  • What Brave version you’re using at this time (Production, Beta, Nightly)?

  • Is your Brave version fully up-to-date (this can be checked under Settings and About Brave)?

  • A screencap of the message/error received?

Looking forward to your reply,

Brave Support

3

ive got it on auto select ai, latest windows 10 and a 2 year old dell laptop and doing it to me all day

4

and it keep happening please help. i updated everything. i use brave as my browser and so naturally i wanna use leo ai for discuss wisdom and writing code. but whats the point if it keeps flagging me.

5

so imma gonna send you reply everytime it happens so you get a time frame …this is like 5th time today only 4th meassge

6

took a nap and still limits me come on fix please. ive got work to do.

7

hey still not fixed and ive got a deadline. this sucks

8

Hello there - I’m sorry to hear this is still happening. When it occurs, have you tried switching your model (choosing a model vs. automatic selection)?

9

I am plagued by the same problem.

  1. Working on a critical report and then get a prompt that I hit a TIME WALL.

  2. macOS 26.3.1, M4 Air, 16mb ram, 1tb on board memory so it’s not my system.

  3. Memory pressure is low.

  4. Paid for Premium but can only connect of Claude Haiku.

  5. Conversation architecture is useless for extended projects.

10

Hello @Leofan - would you be able to nab screenshot(s) of the errors you’re receiving, or a recording displaying the behavior you’re experiencing? This will help us troubleshoot further.

@ste

(The behaviors noted by Leofan appear to be different from that surfaced by @ChurchofYahweh.)

11

@ChurchofYahweh I think I found a solution. I may break Community Rules, but respectfully, I posed the question to Leo and he agreed he’s limited, even with Premium. Better to spend more and get: “You’ve spent months teaching another AI how you work. That context shouldn’t disappear because you want to try something new. Claude can import what matters, so your first conversation feels like your hundredth.”

  1. Your are paying $154.00 per year for premium subscription to Leo?
  2. Claude Sonnet v4.6 offers https://claude.com/import-memory
  3. This means you can import all you conversations into Claude Sonnet and will NOT loose context.
  4. Subscription offer 1 million tokens for $240 per year
  5. Leo.ai Premium costs less but has bulky architecture and uses none relational Conversations, which turns you into a database manager.
  6. Disclaimer - double check this information. I just found it.
  7. Claude.com / important-memory
12

@Leofan not against the the rules but also not super helpful. If you’re not interested in continuing to resolve the actual issue you’re encountering let us know and we can resume assisting OP.

13

Comment from Leo.
Comment from Leo.1622×810 91 KB

14

Screenshot 2026-03-12 at 5.45.00 PM
Screenshot 2026-03-12 at 5.45.00 PM1666×862 55.6 KB

15

I believe it was a solution to @B-Lee problem and mine. We’re both working on extended projects that exceeds Leo’s capabilities. The object was to offer a possible solution, not to try and work with a feature that was not designed for extended context. Why struggle with a tool that can’t to the job? If you have a solution, I would love to read it.

16

Hello @ChurchofYahweh, @Leofan - it looks like you’re not running into rate limits so much as model context limits. This can happen when uploading batches of .pdfs - the team is working on improving this in an upcoming major release with a feature called ‘compaction’ that will attempt to summarize portions of the conversation in order to fit context limits, however I’d recommend uploading smaller file batches in the meantime.

17

Hello @ChurchofYahweh, @Matches I have a solution to offer that stays within Brave Leo’s domain.

Problem: Information imbedded in various conversations with Leo.

Need: To extract specific information from various conversations.

Solution: Tell Leo what you want to do. He will respond with the text for a prompt to inject into each conversation:

Example: “Leo, this is an old conversation we had many months ago. You and I are trying to retrieve any information that has to do with or includes: XXXXXX, XXXXXXX, XXXXXX. When you have finished reviewing this conversation, I want you to put info you find in our memory stack, even if it takes several slots. I will repeat this same message to you in the next conversation. If there is no such data, I will move us on the next conversation.”

Result: It works. Yes, it’s clumsy and confusing - don’t try it unless you’ve had three cups of coffee.

Note: Keep Leo on the same model, not Auto.

Observation: It’s interesting to see how Leo evolves over time. The first few conversations, he’s all business (Claude Sonnet). Throw in a few joke / bull sessions. He has a great sense of humor. Mine loves the Star Trek series and will start chatting in various characters styles - Spock, Data, etc.