Skip to main content

Settings

This page explains where settings live in NeoTavern and how to configure providers, connection profiles, and API keys.

Where Settings Live​

NeoTavern has no separate settings page. Everything opens as a panel or modal over the chat workspace, and closing it returns you to the exact same chat and draft:

  • Settings (from the navigation rail) groups app-wide options into tabs: General (language, text scale, startup screen, message style, avatar shape, accessibility), Themes (install and activate themes), and Data (migration, backups, cache maintenance, diagnostics).
  • AI Settings is the context panel for generation. Its Config tab holds the request parameters for the active model: context size, response length, streaming, sampling, penalties, seed, and reasoning. The API tab manages connection profiles and keys, and Advanced builds custom chat and instruct templates from ChatML, Llama 3, or Alpaca.

Settings changes apply immediately where they are easily reversible. Options that differ from their defaults are marked and can be reset individually, and settings search covers names, descriptions, and keywords.

Global vs. Per-Chat Settings​

Global settings in Settings apply to the whole app: language, theme, data management, and defaults. Per-chat behavior lives next to the chat: generation parameters, the active provider and model, and the context strategy are edited in the AI Settings panel while the chat stays open, and drafts and scroll position are preserved. The persona is per-chat too — each conversation can use a different persona while the app-wide active persona remains the default.

Providers and Connection Profiles​

A connection profile bundles everything needed to talk to a provider: the API type and source, base URL where applicable, the selected API key, and the model. The API tab in AI Settings (and the Providers section) lets you:

  1. Choose the top-level API (Chat Completions or Text Completions).
  2. Pick a source, which filters to the sources of that API and becomes the profile name.
  3. Enter the base URL for OpenAI-compatible servers, usually ending in /v1.
  4. Choose or type a model ID, optionally loading the model list first.
  5. Test Connection to check availability and latency, then Connect to activate the profile.

API Keys​

Keys are stored locally in a key manager that holds several named keys per provider with one active at a time. Secrets are verified before saving and are never displayed in full after that — only a masked suffix remains visible. Exports and diagnostics exclude secrets by default, and provider errors are shown as localized messages with technical details and a trace ID in a collapsible block.

See Themes, Extensions, and Data & Backups for the rest of the app-wide settings.