Settings
You do not need to change anything here: itokens looks after itself. Settings is for when you want to see or choose. Open it at the bottom of the side bar.
Hardware
Hardware shows everything itokens knows about your Mac: the chip with its cores, graphics, memory (and how much is in use now), storage, power, displays, macOS and the AI tools. Rescan hardware looks again.
How your Mac is rated
AI capability shows your Mac (its own picture, from macOS), what itokens measured (chip, graphics, memory and how much of it the GPU may use), and the six levels from Incapable to Highly capable: what each needs, what to expect, and the model a Mac at that level runs, with yours marked.
Models folder
Where itokens keeps its AI models (~/Models unless you change it). Change moves them to another folder or an external disk; anything else in the old folder stays. Keep Quick, Balanced and Deeper ready downloads all three in the background so switching is quick; it skips a model when your disk would have less than 20 GB free.
Which models itokens may use
By default itokens uses models made in the United States (Llama by Meta, Gemma by Google). Each country's model makers are listed with its flag; switch one on to let itokens use its models too. Community versions are models changed by independent people rather than their makers; they are off unless you switch them on. Model details (on unless you turn it off) shows each model's name and maker; off, you see only what each is for (Quick, Balanced, Deeper).
Updates
The first row shows the version you have. Check for updates looks for a newer one; when there is one, Update to installs it (see Updates).
Read links in messages
On unless you turn it off. When your message has a link to a web page (up to three), itokens reads the page and gives it to your AI with your message, and the answer shows what was read. The itokens app reads pages as a browser would, with its own sign-ins, apart from Safari's. When a page needs you to sign in, the answer says so and shows Sign in: sign in in the window that opens, press Done, then Try again. itokens reads only links you write, never links inside pages or files.
Search the web
Your AI can search the web with your own Brave Search API key: get one from Brave's dashboard, paste it in, and press Save. itokens tries the key with Brave first, then keeps it in your Mac's Keychain (not in backups; Remove key takes it out). From then on, when a question needs current information, your AI searches, and the answer shows what it searched for. Only the search words go to Brave; your chats stay on your Mac.
Long chats
Summarize long chats sets when itokens summarizes the earlier part of a chat to make room: at 75% of what your AI can hold (the default) or at 80%. The whole chat stays on your screen either way.
New models
itokens looks for new models every day. When a newer model that fits your Mac comes out from the same maker, itokens downloads it in the background and tests it on your Mac when you are not chatting: it asks it four questions and measures its speed. If it passes, it takes the older model's place. If a model's maker stops offering it, itokens stops using it; when it was the one you use, itokens moves you to a replacement by itself and tells you. Check now looks straight away, and Test tries a model on your Mac. Other models that fit this Mac lists the rest.
Every model in the list shows Test (or Details once tested). It opens a panel with the model's size, maker and what it is, and tests it on your Mac: itokens asks it four questions and shows each answer and whether it was right, and how fast it answered. A model that does not stop at the end of its answers does not pass. Downloading one to test it asks first, with its size.
Use it for is Auto by default: itokens decides whether a model that passed is a good fit for one of your places. You can choose a place yourself (Quick, Balanced or Deeper) for a model that passed; itokens asks first, and if you are not sure, leave it on Auto.
Backup and restore
Back up saves your chats, projects, skills, Library, connections and settings in one file on your Desktop, named itokens-backup- and the date. Models are not in it: they download again. The file is private to your Mac account; keep it somewhere safe.
Keys and tokens in your connections are kept in your Mac's Keychain, not in the backup: after a restore, the Connections page asks for them once. Backups made by itokens 0.2.0 still hold those keys; delete old backup files you no longer need.
Restore replaces your chats, projects, skills and Library with the backup's. After installing itokens again, or after a factory reset, the first screen offers Restore from a backup: itokens then sets itself up with the name the backup had.
Factory reset
Removes everything itokens added to your Mac, then itokens itself, as if it had never been installed. See Uninstall.