FAQ

Yes. Your company can run OpenLoc itself for free forever. Managed hosting and enterprise support are paid options.

ModelFront, an AI company. ModelFront is paid for AI outcomes, not seats or words, so its interest is in AI translation working well, not in selling a platform. That's why OpenLoc is free and open to any AI.

Yes. OpenLoc was shaped with Fortune 500 localization teams as design partners, with SSO, roles and a record of every change. Run it yourself, or get enterprise hosting and support from ModelFront, with an SLA.

Yes. If OpenLoc has to run in your own environment, for example for defence or regulated data, ModelFront can support it there. We work it out case by case. Contact us.

OpenLoc is built for enterprise localization from day one: shaped with Fortune 500 teams as design partners, with SSO, roles, a record of every change, and enterprise hosting and support with an SLA. It is backed by ModelFront, an AI company paid for AI outcomes, not seats or words, so there is a funded maintainer with a business reason to keep it current. And it is AI-native: any model, no AI tax, prompts with context and RAG, and system-wide MCP.

OpenLoc is model-neutral. Choose models by language and task, and change them without changing your TMS.

Connectors in OpenLoc are plain code: free to use, not tied to a plan, and anyone can add or extend one. ModelFront builds the connectors Early Access Program members need, as part of the program. Every connector built that way goes into the shared catalogue, so each team that follows starts with more. See the integrations page for the current list.

The Business Source License 1.1 (BSL). You can read all the source code, and your company can always run it itself for free forever. The full license text ships with the repo on October 19, and the licenses page has the details.

Yes, ask us. We consider other licenses and Additional Use Grants case by case.

The source code goes public on GitHub on October 19, 2026, with a quick-start guide for running OpenLoc yourself. If you'd rather have our engineers on it, book a demo or apply for the Early Access Program. Places are limited, and each team gets free hosting, AI credits, integrations built for them and help with migration.

Anyone can apply. Places are limited, so we prioritise enterprise localization teams where we can make the biggest difference: lots of raw MT to check and fix, a lock-in problem to solve, and ready to move. If you're not in the first group, the code is public from October 19 for running it yourself, and general availability follows once the first teams are live. We'll announce the date here.

MCP stands for Model Context Protocol. It gives AI assistants a standard way to connect to tools and data. With OpenLoc, compatible assistants can use it to scope a release or check translation quality.