Getting started
Your first model
Open the catalog, read one verdict per row, and let the runtime size the model to the memory you actually have.
The hard question about any model is whether it runs on your machine, and at what quality. Conifer answers on the card, before you download. If you have not installed yet, start with Install.
Open the catalog
The studio’s model browser opens on a short list of vetted everyday picks, with specialists below. When you know the shape of your work, the choosing guide routes you to a recommendation by task.
Every row carries three numbers (parameters, quantization, size on disk) and one verdict, and the verdict is the part you read first. Decode speed and intelligence scores live on the models ledger.
Read the fit verdict
Conifer probes your hardware once, derives a usable-memory ceiling from the GPU and the memory actually free, and compares every model against it. Each row gets a small pill, answered before download:
- fits
- Runs with comfortable headroom.
- tight
- Fits, with little room to spare; long contexts may push it.
- won’t fit
- Larger than this machine’s usable memory.
- fit unknown
- The ceiling couldn’t be read, so Conifer says so.
Hover a verdict to see the gigabyte arithmetic behind it.
Download it
Click download and the runtime starts the job; the row swaps its verdict for a progress meter, and the download keeps going while you work elsewhere.
What “fits it to your memory” means
You never chose a quantization or wrote a config file. Conifer picks the precision that lands without swapping weights to disk and sizes the KV cache to the memory free at load time. That is how a 24B model runs on a 36 GB machine.
A few good starting points
These are starting points, not a ranking; the catalog’s fit pill is the authority on your hardware.
| Usable memory | Reach for | Why |
|---|---|---|
| 8 GB | a 3B to 4B dense model | fast, fits with headroom, good for chat and short code |
| 16 GB | a 7B to 8B dense model | the everyday sweet spot for quality and speed |
| 32 GB and up | a 14B+ dense model or a sparse MoE | more capability; an MoE trades RAM for a small model's speed |
Numbers are guidance, not a guarantee. The catalog reads your real ceiling and tells you per model.
Whether a model is dense or a sparse MoE changes how it spends that memory, so the defaults weigh architecture, not parameter count alone. The next step is your first chat.