WyvernChat custom provider setup (AI Connections)

• A WyvernChat account at app.wyvern.chat. The free tier takes your own provider.
• A RelaySwitch key from the console (email code sign-in). Keys start with sk-relay-.
• A balance. Top-ups start at $10, by card or crypto, with no subscription. Before the first top-up the console chat runs 12 free turns a day.

Open AI Connections, stay on Providers, click + Create Provider and fill the form:

• Chat Completion Only? on: the API answers the chat completions format; with it off Wyvern may try a text-completions route and get 404.
• No Key Required off, and the key in Stored API Key with nothing in front of it. If your release shows the key field only after saving, save, then Add API Key on the provider's row.
• API URL: https://api.relayswitch.io/v1 with the /v1. If Fetch Models or the first message answers 404, your release posts to the address as typed: use https://api.relayswitch.io/v1/chat/completions instead.
• Allowed Parameters: leave empty. Click Add Provider.

Switch to Connections, click + New Connection, name it after the model, set Visibility Private, and under Model Configuration:

• Fetch Models works here: the API answers /v1/models with your key, so the list fills with every id on the Models page. If it fails, type the id exactly; a display name answers 404 model_not_found.
• Context Length (tokens) is the number that sets the bill; the table below gives a value per model.
• Tool Call Enabled, Inline Image Support and Use Browser Request: off. With browser requests off, Wyvern's server makes the call and the key never reaches the browser.
• Pick the Stable preset under Parameters; the uncensored builds need no jailbreak in the instruct template. Click Create.

Then, in any chat: the cog icon, Connection Settings, Free Queue off, your connection under Mine. A second model is a second connection on the same provider.

Wyvern's Free Queue serves whatever is free that minute, with a queue in front of it; your own connection serves the model you picked, at once. Prices are per million tokens; the turn columns price one 2K-in / 300-out turn: 2,000 tokens of card, persona and recent history in, 300 out.

The flagships, for longer memory and stronger writing:

Every message resends the character, persona, lorebook entries and as much history as Context Length allows. Per message at three sizes, with a 300-token reply:

Set Context Length at or under the model's window in the Context column, never above it: a prompt past the window is refused, not trimmed. The console's Usage page lists every request with its tokens and charge.

• 401: the stored key is wrong or revoked. Edit the provider and paste it again.
• 404 route_not_found: the API URL has no /v1, or Chat Completion Only? is off; if both are right, use https://api.relayswitch.io/v1/chat/completions. model_not_found: a typed display name; Fetch Models again.
• 402: the balance is empty or the key reached its budget.
• "This model is having trouble responding": a sampler preset the model rejects, or a reply cut inside a reasoning step. Pick Stable; give a reasoning model 1,000 or more response tokens.
• The reply ignores the character: still on the Free Queue. Open Connection Settings in the chat and pick your connection.

Wyvern's own guide page (read 2026-11-02) sends new users to AI Connections to add a provider and a connection; the field labels are the February 2026 walkthrough's, since the page itself sits behind a sign-in. Wyvern is in beta and renames fields between releases; match each by what it asks for.