No single model is best at everything.
Connect the providers you already use, from Anthropic, OpenAI and Google to OpenRouter, or run models locally with Ollama, and pay them directly with no markup. Then give each agent the model that suits its job.
Put your strongest model on the hardest decisions, and faster, cheaper ones on routine work and automations.
API keys and sign-ins are locked in your system’s secure keychain, never written to a config file and never sent anywhere but the provider they belong to.
The right model for every job, and the bill to match.
| Anthropic | API key |
| ChatGPT | Sign in with your ChatGPT subscription |
| GitHub Copilot | Sign in with your GitHub account to use your Copilot plan |
| OpenAI | API key, or sign in |
| API key, or sign in | |
| OpenRouter | API key, or sign in |
| DeepSeek | API key |
| Qwen | API key |
| Ollama | On your machine with no key, or Ollama Cloud with one |
| Your own endpoint | Any server that speaks the OpenAI Chat Completions format |
| Per agent | Every agent can use its own model, and a crew runs on the worker model you pick · agents · crew |
| Per project | Each project keeps its own model |
| Automations | A small decision model, local or hosted, decides whether to start an agent · automations |
| Where they live | Keychain on macOS, Credential Manager on Windows, Secret Service on Linux |
| Where they don't | Never in a plain-text file or an environment variable where any script could read them |
$10 a month after a 14-day trial · Questions