Bring your own key
Plug in your own API key and pay your provider directly — your rates, your limits, your relationship. alexandr never sits between you and your model bill.
Any model.
Your terms.
alexandr is model-agnostic by design — bring your own API key, run a model on your own hardware, or switch providers whenever a better one ships. Your workspace doesn’t move.
- Bring your own key
- Local models
- Every major provider
- Switch any time
Works with the models you already trust
- Anthropic
- OpenAI
- Mistral
- Local / self-hosted
Three ways in
However you want to run a model — your key, your machine, or whatever ships next — the workspace is the same.
Run local models
Point the assistant at a model running on your own hardware. No key, no cloud, no data leaving the machine — the same workspace, fully offline-capable.
Works with every provider
Anthropic, OpenAI, Google, Mistral and any OpenAI-compatible endpoint. New provider tomorrow? Add it — your apps, skills and history don't care which model answers.
What model-agnostic means in practice
Switch mid-conversation
Change the model from the composer, whenever you like. Draft with a fast model, finish with a strong one — the conversation carries over.
Your workspace stays put
Apps, skills, routines, files and history belong to your workspace, not to a provider. Switching models moves nothing and breaks nothing.
No markup, no middleman
With your own key, tokens are billed by your provider at their prices. alexandr charges for the workspace, never for your model usage.
Right model for the job
Use a small local model for quick edits and a frontier model for hard problems. The choice is per-conversation, not per-platform.
Nothing to migrate
There's no export day. Because nothing is tied to one provider, trying a new model is a dropdown, not a migration project.
Private by placement
Pair a local model with a workspace running on your own Mac or on-premise, and your prompts and data never cross your boundary at all.
The model is one half of the story — the assistant is what puts it to work across your files, apps and routines.