Translation Engine - AI Providers

Full LLM translation - follows the translation prompt, the website description, the glossary and the reasoning effort setting. Billed per token.

- Get an API key: https://platform.openai.com/api-keys
- Step-by-step guide: https://help.openai.com/en/articles/4936850-where-do-i-find-my-openai-api-key
- Key format: starts with `sk-`

Recommended mode: Background. It is the only provider offering it, and it never depends on your server surviving a long request. If speed does not matter, use Batch instead - same result for about 50% less.

Full LLM translation, same prompt/context/glossary support as OpenAI. Billed per token.

- Get an API key: https://console.anthropic.com/settings/keys
- Step-by-step guide: https://platform.claude.com/docs/en/get-api-key
- Key format: starts with `sk-ant-`

Claude has no background mode, so asynchronous jobs always go through the Batch API - and Claude batches usually take 2–8 hours.

Full LLM translation, same prompt/context/glossary support. Billed per token.

- Get an API key: https://aistudio.google.com/apikey
- Step-by-step guide: https://ai.google.dev/gemini-api/docs/api-key

A machine-translation engine, not an LLM. It translates text instead of following instructions, so the translation prompt, the website description and the reasoning effort are ignored. Glossary terms are respected - they are uploaded to DeepL as a native glossary. Billed per character.

- Get an API key: https://www.deepl.com/your-account/keys
- Step-by-step guide: https://developers.deepl.com/docs/getting-started/auth
- Free keys end with `:fx` and are routed to the free endpoint automatically.

Synchronous only (no Batch API), but content is split into many small requests, so long posts are safe.

A machine-translation engine, not an LLM. The translation prompt, website description, reasoning effort and the glossary are all ignored. Billed per character.

- Get an API key: https://console.cloud.google.com/apis/credentials
- Step-by-step guide: https://cloud.google.com/translate/docs/setup
- Requires a Google Cloud API key with the Cloud Translation API enabled.

Synchronous only, content split into small requests - safe for long posts.

The processing mode decides how each translation is sent to the provider and when the result comes back.

The cron calls the provider and waits for the answer in the same run, so the translation appears within a minute or two. Full price.

Best for short content (terms, titles, strings) and for seeing results immediately while setting things up.

The request is handed to the provider's Batch API for about 50% of the normal price and polled every minute until it is ready.

- OpenAI / Gemini: most batches come back in 10–60 minutes (typically 20–30).
- Anthropic (Claude): usually 2–8 hours.

The only guarantee providers give is a 24-hour window, so a single job can take much longer at peak times. Best for bulk translating many or long posts, where cost matters more than speed.

The request is started without blocking and polled every minute, usually finishing in a few minutes. Full price, but it never depends on your server surviving a long request - the safest choice for long posts.

One long post is a single HTTP request that can take minutes, so it may be cut short by a PHP, web-server or proxy timeout - and the whole translation is then retried from scratch. For long posts and bulk jobs pick Batch (or Background on OpenAI).