Skip to main content
byok-models

Connect to powerful large language models using your own API keys — giving you full control, privacy, and flexibility. AI Thing never sends data to any model provider unless you choose to connect your own key.
Now you can use multiple models in the same conversation.

​
BYOK: Bring Your Own Key

You choose the model. You bring the API key. You stay in control.
AI Thing does not charge for usage — you pay only your provider, directly. AI Thing now supports Anthropic Claude, OpenAI GPT-5, and Google Gemini 2.5 / 3, plus more coming soon.

​
Supported Models

​
Anthropic Claude (4.5 series)

  • claude-haiku — Fast and cost-effective
  • claude-sonnet — Balanced
  • claude-opus — High intelligence
Create API Key: https://console.anthropic.com/settings/keys

​
OpenAI GPT-5 Series

  • gpt-5-nano — Lightweight, cheap
  • gpt-5-mini — Fast and capable
  • gpt-5.1 — Advanced reasoning and long-context
Create API Key: https://platform.openai.com/settings/organization/api-keys

​
Google Gemini

  • gemini-2.5-flash — Extremely fast, low-cost
  • gemini-2.5-pro — Strong all-purpose reasoning
  • gemini-3-pro — Next-generation performance
Create API Key: https://aistudio.google.com/app/api-keys

​
How to Connect

  1. Get an API key from your chosen provider
  2. Paste the key into the corresponding field inside Settings → Models
  3. Select the default model you want AI Thing to use
Keys usually look like:
  • Anthropic: sk-ant-...
  • OpenAI: sk-...
  • Google: AIza...
Your keys are stored locally on your device, never uploaded anywhere.

​
Add Your Own Models

The open-source version of AI Thing includes a Models.plist file inside the project.
You can add custom models by editing this file:
  • Add a model ID, display name, and provider type
  • Save the file and rebuild or relaunch
  • Your model will appear in Settings → Models and can be used immediately
This makes it easy to plug in local, experimental, or self-hosted models with minimal setup.

​
Model Settings

Fine-tune how each model behaves in Settings → Preferences:

​
Prompt Cache

Reuses common parts of a prompt to speed up responses and reduce cost during long or multi-turn conversations.
Best used when you’re working in the same thread for a while.
Learn more

​
Output Tokens

Set the maximum possible response length — helpful for keeping outputs concise or allowing long explanations.
Don’t want to manage API keys? Email us at [email protected].
Have questions? Email us at [email protected].