For the complete documentation index, see llms.txt. This page is also available as Markdown.

SillyTavern

Overview

SillyTavern is a locally installed AI chat interface for creating and interacting with custom AI characters. This guide shows you how to integrate MegaNova's API as a backend for SillyTavern using OpenAI-compatible endpoints.

Key Features

  • Free tier available for supported models

  • OpenAI-compatible API endpoints

  • Multiple model options including Manta, DeepSeek, Mistral, and Nevoria

  • Local installation ensures privacy

  • No external hosting required

Requirements

  • MegaNova account with API access

  • SillyTavern - Download and install

  • API key from MegaNova

Usage Policy: All integrations must comply with our Terms of Service and Privacy Policy, including acceptable use policies and content guidelines.

Integration Steps

1. Get Your API Key

MegaNova API Keys Dashboard
  1. Log into your MegaNova AI and navigate to API Keys Page

  2. Copy your default API key and securely store it

2. Install SillyTavern

Download and install SillyTavern following the instructions here: https://docs.sillytavern.app/installation/

3. Configure SillyTavern API

In the top menu bar, click Configure to open the API settings panel. You'll see options for different backends.

Configurate settings
  • Under Chat Completion Source, select Custom (OpenAI-compatible).

  • In Custom Endpoint (Base URL), enter:

  • Leave /v1 at the end — SillyTavern will append /chat/completions automatically.

  • Paste your API key into Custom API Key field.

4. Choose a Model

MegaNova offers a wide range of models suited to SillyTavern's use cases:

  • Manta-Mini-1.0

  • Nevoria-70B

  • Mistral-Small-24B

  • DeepSeek

  • And many more

In the Enter a Model ID field, paste your chosen model exactly as listed in MegaNova's serverless models:

Choose model

5. Connect and Start Roleplaying

Connect proxy
  • Once the status shows Valid (green dot), you're ready to go.

  • Click Connect in SillyTavern.

Your SillyTavern instance will now use MegaNova's API for AI model access.

chat with character

Available Models

The following models are compatible with SillyTavern:

Free Models (No Payment Required)

Model
Description
Best For
Pricing

manta-mini-1.0

Lightweight dialogue-optimized model

Quick responses, casual chat

Free

L3.3-MS-Nevoria-70B

70B parameter model for advanced reasoning

Long-form dialogue, complex reasoning

Free

L3-70B-Euryale-v2.1

70B parameter creative writing model

Creative writing, storytelling

Free

L3-8B-Stheno-v3.2

8B parameter efficient model

General tasks, efficient generation

Free

DeepSeek-V3-0324

685B parameter model for advanced reasoning

Advanced reasoning, complex tasks

Free

Free Model Usage: Free models have daily limits and are intended for evaluation and interactive use only, not production workloads.

Model
Description
Best For
Pricing

manta-flash-1.0

Balanced tier for general-purpose use

General conversations, balanced performance

$0.02/1M in, $0.16/1M out

manta-pro-1.0

Advanced tier with long context support

Complex dialogue, detailed interactions

$0.06/1M in, $0.50/1M out

nevoria-70b

70B narrative generation, LoRA tuning & story-rich outputs

Creative storytelling, character depth

Paid

mistral-small-24b

24B instruction model, long context, fewer errors

Natural conversation flow

Paid

Character Configuration

When creating or editing characters in SillyTavern:

  1. Set API Provider to "OpenAI"

  2. Configure Temperature (0.7-0.9 recommended for roleplay)

  3. Set Max Tokens (2048-4096 recommended)

  4. Add character-specific prompts in the System Message

Advanced Configuration

Custom System Prompts

Temperature Settings

  • 0.7-0.8: Balanced creativity and consistency

  • 0.8-0.9: More creative and varied responses

  • 0.9-1.0: Maximum creativity (may be less coherent)

Configuration Tips

  • Set temperature between 0.7–0.9 for creative responses

  • Maintain chat logs for character continuity

  • Experiment with different system prompts to adjust character behavior

  • Example system prompt:

"You are a sentient being who remembers past interactions, speaks naturally, and adapts emotionally."

Performance Optimization

  1. Model Selection: Choose the right model for your use case

  2. Token Management: Optimize prompt length to stay within limits

  3. Caching: Use SillyTavern's caching features for better performance

  4. Batch Processing: Group similar requests when possible

Support

Conclusion

This integration allows you to use MegaNova's API models with SillyTavern through OpenAI-compatible endpoints. For additional support or questions, refer to the resources listed in the documentation.

Last updated