The BYOK & Local-LLM Interview Copilot Explained

A BYOK (bring-your-own-key) interview copilot is a tool that, instead of using its own closed cloud model, uses your API key or a local model on your device. That gives you the same AI power as cloud competitors — without their subscription, and without sending your data through a middleman.

The Three Modes

Natively is the only assistant that supports all three modes in one app:

• Local (Ollama): local models (Llama, Mistral, CodeLlama, Gemma) + local Whisper — 100% offline, zero traffic, zero cost.
• Cloud (BYOK): your own OpenAI, Anthropic, Gemini, or Groq key — or any OpenAI-compatible endpoint. The same power as cloud competitors, pay only for tokens.
• Managed (Natively API): hosted STT + LLM on a subscription, no keys needed.

Why BYOK Beats a Subscription

Cloud copilots charge a fixed monthly fee regardless of how much you use them. With BYOK, you pay the model provider only for the tokens you actually use — often cents per interview session. For most candidates that is significantly cheaper than a $25–70/mo subscription.

For a full price comparison, see our breakdown of interview copilot pricing.

Which Models and Providers Are Supported

• Google Gemini: 3.1 series (recommended; up to 2M-token context).
• OpenAI: GPT-5.4, o3 series.
• Anthropic: Claude 4.6 series.
• Groq: Llama 3.3 (text), Llama 4 Scout (OCR/vision).
• Ollama / LocalAI: Llama, Mistral, CodeLlama, Gemma — fully offline.
• OpenAI-compatible endpoints: vLLM, LM Studio, OpenRouter, DeepSeek, and custom.

The Privacy Advantage of BYOK and Local

In local mode, your data never leaves your device. In BYOK mode, data goes directly to the provider you chose under your own key — with no intermediate copilot server storing it. That removes an entire class of privacy risk inherent to closed cloud tools. More in our article on AI interview assistant privacy.