Feature
Bring your own AI model into chat
SayPilot is the floating reply layer. You choose the model, provider, quality level, and cost profile that fit your needs
- DeepSeek
- GPT
- Claude
- Gemini
- Qwen
- Custom API
One reply workflow, many providers
Keep the same floating assistant experience while routing generation to the model setup you prefer
Control quality and budget
Different chats may need different models. SayPilot keeps provider choice visible and manageable
FAQ
Questions this page answers
Can I use an OpenAI-compatible endpoint?
Yes. SayPilot is designed to support custom OpenAI-compatible model endpoints.
Which advanced model parameters can I set?
For OpenAI-compatible providers, the useful controls are model name, base URL, API key, temperature, max tokens, system prompt, and provider-specific options when the app exposes them.
How should I tune advanced parameters?
Start with the provider default. Lower temperature for steadier replies, raise it for more varied wording, use max tokens to limit reply length, and adjust the system prompt only when you need a consistent rule or style.
Do custom models use SayPilot cloud credits?
No. When SayPilot routes generation to your configured provider, it does not deduct SayPilot cloud reply credits. The model call consumes your provider tokens or quota. SayPilot may still require an active subscription to enable custom-provider routing.
Do I need a subscription or a credit pack?
For the planned paid model, bring-your-own-provider is a subscription feature, not a credit-pack flow. The subscription unlocks custom-provider routing; actual reply generation then uses your provider tokens or quota. SayPilot credits still apply when you use the default SayPilot cloud model.
Does model choice change the app flow?
No. Model selection is part of the same reply generation workflow.