← Back to Home

Best Janitor AI Kobold Settings for Speed

Many users try to use Kobold AI with Janitor AI to avoid paying for OpenAI API keys. However, Kobold often suffers from extremely slow generation times and connection timeouts.

Kobold AI Issues

  • Requires running code on Colab
  • Slow response (2-5 mins)
  • Disconnects frequently

The Better Way

  • Candy.ai (No Setup)
  • Instant text generation
  • No technical knowledge needed

Recommended Kobold Presets

If you insist on using Kobold, try these settings to reduce lag:

  • Preset: Godlike or Erebus
  • Context Size: Lower to 2048 (Higher = Slower)
  • Temperature: 0.7
  • Repetition Penalty: 1.1

Stop Waiting for Text to Load

Configuring endpoints and Google Colab links is a headache. Why not use a platform that hosts the models for you?

Candy.ai runs on dedicated high-speed servers. You just log in and chat. No API URL, no Kobold, no waiting.