Your keys. Your models. Your code.
Koadr is model-agnostic and BYOK. No lock-in, switch models per task, and your keys never leave your control.
OpenRouter, Anthropic, OpenAI and Google are enabled by default.
Cloud
22- Anthropic
- OpenAI
- DeepSeek
- Groq
- xAI
- Mistral
- OpenRouter
- Together AI
- Fireworks
- Cohere
- Perplexity
- Cerebras
- SambaNova
- Hyperbolic
- Novita
- DeepInfra
- Azure OpenAI
- AWS Bedrock
- Vertex AI
- GitHub Models
- Requesty
Local
2- Ollama
- LM Studio
Settings · providers
BYOK- OpenRouter
- Anthropic
- OpenAI
- DeepSeek
- Groq
- Mistral
- xAI
- Ollama
- LM Studio
Keys are stored per workspace. Nothing is sent anywhere you did not configure.
FAQ
Questions, answered
Frequently asked questions
- Yes. Koadr is BYOK: you connect your provider accounts and pay them directly. We never mark up model usage.
- Yes. Point Koadr at Ollama or LM Studio on your machine. If the endpoint is reachable, Koadr uses it like any cloud provider.
- Yes. The model is a per-task choice, not a project setting. Different steps of the same build can use different providers.