Work with long documents in a single conversation
A larger model and a 131,000-token context window let Timberline read entire manuals and policies at once instead of fragments.
Timberline now runs on a larger model with a context window of about 131,000 tokens, and it runs privately, like every model in Timberline.
In practice, that means long material can be read in full rather than in pieces. A 134-page manual fits in a single conversation. When a document is larger than the window, Timberline keeps the pages most relevant to your question instead of cutting it off arbitrarily.
Follow-up questions about the same document are also faster, because the document doesn’t need to be processed again for each question.