TinyOffice AI
TinyOffice AI is the default. Sign in with your TinyOffice account and it works — no API key to find or paste. Usage is paid for with the credits that come with your plan. The list of models is kept up to date from the server, so new models appear without an app update.
Custom providers
TinyOffice can also talk to AI providers directly with your own API key. Custom providers are off by default. When they're available in your build, you'll find them under AI Model → Enable custom AI providers, with Provider, Model, API Key, Base URL, Test connection and Max output tokens.
Supported providers with an API key:
| Claude | OpenAI | Gemini |
| DeepSeek | Qwen | Mistral |
| Grok | Kimi | GLM |
| MiniMax | Doubao | OpenRouter |
| OpenCode Zen | OpenCode Go | LLMCloud |
TinyOffice can also drive coding-agent CLIs installed on your computer — Claude Code, Codex, OpenCode CLI, Pi, Droid and Cursor — and any OpenAI-compatible endpoint through Custom, including local servers such as Ollama, LM Studio or vLLM (the API key is optional there).
Where your key goes
A custom API key is stored only on this device. Requests go straight from TinyOffice to the provider you chose.
Cloud tools — web search, image generation, media analysis — always run through TinyOffice, so they still need you to be signed in even when you use your own key. When cloud tools are off, image tools are unavailable and web search uses free sources.