Documentation
Advanced providers, when you need them.
Capper’s normal setup uses built-in local processing. External providers are optional, and existing choices are retained on upgrades.
Bionic / LM Studio
For an existing external setup, these providers use http://127.0.0.1:1234/v1. Enable Bionic’s Local Model API, or load a chat model and start the server in LM Studio. Select the matching provider in Capper’s Advanced settings.
Ollama
For optional Ollama use, start its server with a chat model loaded. Capper expects http://127.0.0.1:11434/v1. Follow Ollama’s own installation and model instructions.
Custom servers
Enter the base URL of a compatible /models and /chat/completions API in Settings. A non-loopback URL sends your instruction and captured text to that server. Capper cannot verify its downstream handling.
Check an external connection
Your external host manages its own models and server lifecycle. Keep it running with a chat model loaded, select the matching preset, and use Test connection. If the list is empty, use Refresh Models or Apply & Refresh after editing a custom URL. Model behavior depends on your external setup and hardware.
A valid paid license or active local trial is still required for local requests. Optional larger downloads and custom styles are Advanced choices, not first-run requirements.
Processing labels and privacy
On this Mac identifies Capper’s embedded text and local speech processing. Separate local servers and custom servers have their own labels; they are not equivalent to embedded processing. Setup downloads, license validation, and optional update checks can use the internet. Read the privacy details.
Provider references
For current external-server controls, see the LM Studio server guide and Ollama compatibility documentation.