Model neutrality pledge

Why neutrality matters

Today, most AI products are a front end for one company's models. That's fine until a better model ships somewhere else, prices change, or a provider's policies stop fitting your business. Then everything you've built — your prompts, workflows and history — is stuck where it is.

alexandr is designed the other way around. The workspace is the permanent part. The model is a part you can swap.

What we commit to

• Multiple providers, always. alexandr will support models from more than one independent provider, including models you can run yourself.
• Switching is free. We won't charge you to change providers, and we won't make switching harder through the product.
• Your work is model-independent. Your apps, skills, connectors, automations and history keep working when you change models.
• No paid placement. We won't take money from a model provider to make their model the default, or to rank it above others.
• Honest comparisons. When we recommend a model for a task, it's because we think it's the best fit — and we'll tell you why.

What neutrality doesn't mean

Neutral doesn't mean every model behaves the same, or that we'll support every model ever released. We add providers when their models are good, safe and reliable enough for real work. And each provider's own usage policies still apply to the requests sent to it.

Keeping ourselves honest

We publish the list of supported providers in alexandr itself, and any change to this pledge will be announced in our news before it takes effect.