Free. Open source.
Self-hosted.
Apache-2.0. The runtime is the product; we don't gate features behind a paid plan because there is no paid plan. Run it on your hardware, fork it, ship it. The promise of local-first is the whole point.
What you pay for.
The runtime is free. Two things are not, and they're not ours to sell:
- LLM inference. Whatever you pay your chosen provider — OpenAI tokens, Anthropic credits, OpenRouter credits — flows through your account, not ours.
- The hardware. Your laptop, your server, your homelab, your AWS instance. We touch none of it.
No tiers, on purpose.
We don't ship a "team" tier with cloud sync, a "business" tier with single sign-on, or an "enterprise" tier with procurement-friendly pricing. The runtime is the same whether one person uses it or a thousand. If your organization needs SSO, you wire it in front of the dashboard yourself; the code is yours to modify.
No hosted version planned.
We're not running a hosted Axocoatl product. Local-first is a brand commitment, not a launch tactic. The promise that nothing leaves your box unless you wire it to is the whole thesis — flipping that later would invalidate the original choice.
If you need help.
GitHub Discussions are the support channel. They're public,
searchable, and free. If you'd rather pay for direct help,
the project maintainers are reachable for paid engagements;
contact details are in SECURITY.md and the
contributor list. That's a side arrangement, not a product.
Use it for free.
The only cost is your time to install and the inference tokens you choose to pay for.