Skip to main content
Yes. Kodus supports Bring Your Own Key (BYOK) on every plan, letting you use your preferred LLM provider. You pay the provider directly — Kodus never marks up tokens and never sees your key in plain text.

​
How it works

  1. In the sidebar, open AI providers (app.kodus.io/byok)
  2. On the Providers tab, pick your provider — or OpenAI-compatible / Anthropic-compatible under Custom for any other endpoint
  3. Paste your API key, choose the model, and click Test & save — Kodus tests the connection before saving

​
Supported providers

The Providers section of the grid lists first-class providers such as Anthropic, OpenAI, Google, OpenRouter, Novita, Moonshot, and Z.ai. The Custom section covers endpoints you point Kodus at yourself: OpenAI-compatible, Anthropic-compatible, Google Vertex AI, Amazon Bedrock, and Azure OpenAI. When Kodus can list a provider’s models, the Model picker shows them (recommended ones first, plus Type model manually); otherwise, you type the exact model ID. Thinking/reasoning, temperature, token limits, and max concurrent requests are under Advanced (optional).

​
Why use your own provider

  • Data control — requests go directly to your provider
  • Cost management — use your existing billing relationship, no markup
  • Model choice — pick the model that works best for your codebase
  • Compliance — meet internal requirements for AI tool usage
  • Fallback resilience — set a Fallback (optional) model on the Routing tab to keep reviews running during provider outages

​
Self-hosted LLM providers

If you self-host Kodus, you can point BYOK at self-hosted or aggregator endpoints like:
  • Novita, Groq, Together AI, Fireworks AI
  • Chutes, Synthetic
  • Your own Ollama / vLLM / TGI instance (via the OpenAI-compatible provider)
For setup guides, see the Cookbook and the provider-specific docs in the knowledge base. For the full BYOK flow (providers, routing, advanced tuning), see Bring Your Own Key.