Immersive Translate
Use RouteMux models for AI translation in Immersive Translate.
Immersive Translate can call any OpenAI-compatible endpoint as its translation engine.
Configure
| Field | Value |
|---|---|
| Translation service | OpenAI (custom / compatible) |
| API endpoint | https://api.routemux.com/v1/chat/completions |
| API key | your RouteMux key (sk-...) |
| Model | a slug from GET /v1/models |
Some versions want the full endpoint path (ending in /chat/completions), others want the
base URL and append it themselves. If the first request 404s, switch between the two forms.
Pick a cheap, fast model
Page translation issues many short, highly parallel requests. Cost and latency dominate here far more than reasoning depth — a small model is usually the right choice, and the per-1M-token comparison on the Models page makes the difference concrete.
Rate limits
Translating a long page fans out into many concurrent calls. If you see 429, lower the
tool's concurrency setting rather than retrying harder — retries against a rate limit make
it worse.
Billing
Only successful translations are billed. Requests that fail or time out cost nothing, so a partially translated page is only charged for the segments that actually came back.