your journal and your computer's memory: local-first, and your cloud options
your journal gets processed. what you've seen and heard goes in as text and images, and audio becomes text too. all of it runs locally, by default. the bundled model runs right in your journal, so your thinking never leaves your device. a cloud lane is there only as an option: for a machine that can't run a local model, or if you'd rather not spend your device's own power on it.
local is the default
out of the box, your journal is processed on your own device:
- thinking and screen-analysis: the reasoning over your journal, and describing what's on screen, run through a bundled local model in your journal. your thinking never leaves your device.
- transcription, turning audio into text, runs locally by default too, with a small on-device model. the one exception is a choice you make: with confidential processing on, transcription can run on the service too. see the transcribe audio on the service setting in the journal's thinking app.
if a computer can't run a local model and you haven't chosen a cloud lane, your journal has no model to think with yet.
what it takes to run locally
the bundled thinking model needs 6 GB of GPU memory on linux and windows, or 16 GB of memory on apple silicon. on disk that takes a few gigabytes, plus the transcription model.
on linux the local model runs on a supported hardware GPU (amd, nvidia, or intel). without one, your journal won't run a local model. windows has the same 6 GB bar on a compatible GPU, and a GPU below it isn't supported. with journal 2.0.64 or later, a windows arm64 pc is checked against the same bar, though local thinking hasn't been tested on an arm64 GPU. see what your machine needs to think locally for what the journal checks, and when.
if your computer is below the bar, that's what the cloud lanes are for.
choosing in the thinking app
on the computer your journal runs on, open your journal and go to its thinking app. it lays out how your journal gets processed as a few lanes, one active at a time:
- local: the bundled model runs right in your journal, so your thinking never leaves. this is the default.
- your own model: your own key from Claude, Gemini, or GPT, your own ChatGPT plan, or your own endpoint URL. your billing, your control; a key stays in your journal. with your own model, your computer talks to the provider or endpoint you chose directly, and it sees the parts of your journal that processing uses. this is a cloud option if your machine can't run local.
- confidential processing: a paid service. processing off your own device, without using your device's own power: your thinking leaves your device and goes to confidential hardware sol pbc runs. no content retained, no human review, nothing used to train. your journal itself stays on your device either way. it's complimentary for approved scouts.
if your journal is using too much memory
- open the thinking app and confirm the local lane is active and your machine has the memory and a supported GPU for it. that's the intended footprint on a capable machine.
- if your machine is below the local bar, switch to a cloud lane: your own model, or confidential processing.
- make sure you're on a current version. see keeping solstone up to date.
- still stuck? see getting help with solstone, or file a request and include your
solstone journal doctoroutput (on a mac, run the command from open admin terminal in the journal app's journal menu).