AI providers and access
WirdDeutsche synchronisiertFassung
TranslatePilot translates through the AI service of your choice. Provider, access and model are configured in the component options (Components → TranslatePilot → Options, tab "Translation").
The three provider variants
platform.claude.com). Billing is per token; TranslatePilot shows a USD cost
estimate before every run and enforces the cost limit.
Mittwald AI Hosting. You create the API key in mStudio, section AI
hosting. The models run on European infrastructure; billing is a monthly flat plan with a token
quota. TranslatePilot therefore shows no USD estimate but how many tokens a run is expected to
take out of the quota.
Other OpenAI-compatible provider. For any service speaking the widespread
OpenAI interface. Additionally enter the base address of the API, including the version path —
for example https://example.org/v1.
Choosing a model — the list comes from the provider
Once provider and API key are saved, TranslatePilot queries the available models directly from the provider and offers them in the model field. You always choose from the current offering, without typing model names.
The list appears only after saving provider and key — save the options once and reopen them. When the provider is temporarily unreachable, the last saved model value stays selectable; nothing gets lost.
Switching providers
Switching is possible at any time: change the provider, enter the matching key, save, pick the model from the freshly loaded list. Two things are deliberate:
Key security
The API key is used server-side only, never output to the browser and never written to logs.