Choose models for all encounters in this workspace.
Not cached · Download needed
Model files not fully cached
Fastest · basic accuracy
Approx. 125 MB download · 500 MB GPU memory
Not cached · Download needed
Model files not fully cached
Better drafting · more memory
Approx. 1250 MB download · 2200 MB GPU memory
Loaded models stay in memory for this app session. Unloading frees memory and keeps downloaded files cached.