What is an LLM?
A large language model is a very large statistical model of text. It does not look anything up unless you give it something to read. It produces the most plausible continuation, which is why it sounds fluent even when it is wrong.
What matters here
- Trained on a large body of text
- Predicts the next piece of text, repeatedly
- Handles many languages and formats
- Has no built in source of truth
- Runs on specialised hardware, usually hosted
What the training buys
The model picks up grammar, tone, common patterns and a great deal of world knowledge compressed into its weights. That is why it can draft a polite email without being taught email specifically.
Where the limits come from
There is no lookup step and no fact database in the base model. If the answer was not in the training data, or the question needs today's information, the model will still produce something confident and possibly false.
Sizes and trade-offs
Larger models tend to handle harder reasoning and follow instructions better, at higher cost and latency. Smaller models are cheaper and faster and fine for classification or short rewrites.
Frequently asked questions
Keep reading
- What is a filter?A filter narrows a list to the items that match a condition, such as a date range, a tag or a status.
- Prysm Note for job huntingA job search is a pipeline, not a to do list: track applications, follow ups and interview prep in one place so nothing goes quiet.
- Prysm Note for writing a bookA book is a long project with a research phase, an outline and a daily word count. Keep all three in one place and the writing stops being a mystery.
Turn talk into a plan
Capture a thought in plain language and let Prysm Note shape it into tasks and a timeline.
Disclaimer
This page is general information, not professional, legal, financial or medical advice. It describes common ways of working and how Prysm Note approaches them. Software changes over time, so details may differ from what you see in the product. Any product, company or brand names that appear are the property of their respective owners and are used only for identification. We do not claim any endorsement or affiliation. Where we compare approaches, we aim to describe each one honestly, including where a different approach may suit you better. Try the tools that matter to you and judge for yourself.