I am curious to know what the context length is for the different subscription plans. I am on the essential which I love for the long conversations but was curious how far it remembers. If I am using Opus for example, do I get the full context, or can you provide how long the context is?
Thanks again! 
Great question! The short answer is it depends on the context of the conversation. The long answer is that since we have more complicated tool use (writing + executing code, troubleshooting error messages, injecting data tables, etc.) than a typical simple chatbot, we don’t have a strict context window of x amount of characters. Instead, we have a dynamic system of including certain parts of the conversation history and truncating/summarizing other sections. This helps with maintaining a fairly long conversation “memory” while also enabling many use cases outside the ability of basic chatting. Functionally, this can look like a context window of 5-10k tokens, but as I said previously, it really depends on the conversation.
The message input length limit is also different than the conversation history length limit. You can go ahead and upload PDFs and larger text files, and we use Opus’ full context length for summarization and chatting with those files. Hope that helps clear things up a bit, but let me know if you have any follow-up questions!
Thank you so much, Zach! Your explanation makes sense to me, and if I were to upgrade to the highest subscription, does that mean I would get a more considerable conversation history? If that gives me even more extensive input lengths, that might be really attractive to me to upgrade my plan.
Keep me posted, but again, I really appreciate your explanation, as that makes much more sense to me.
It might not lengthen the conversation history context window, but it would mean more input tokens (2,500 characters for Basic, 3,000 for Essential and 3,500 for Pro). We are constantly experimenting with this though, and are also working on new ways to lengthen input character limits and improve conversation memory.
This is good to know! I upgraded my plan to the “Team” plan to test out the input length and have been highly impressed. Thank you for your feedback and transparency, as it helps new members understand the input tokens.
Of course! It’s useful to know that’s something we can do a better job of clarifying upfront. I’m sure plenty of others have the same questions