Skip to content
Supervertaler Docs
Searching: All docs

Start typing to search the docs.

Batch Translation

Translate multiple segments at once with AI.

Starting Batch Translation

  1. Select segments to translate:

    • Click first segment, Shift+click last for a range
    • Or use Edit → Select All (Ctrl+A)

  2. Start batch:

    • Go to Translate → Batch Translate and choose which segments to translate, for example Translate selected not-started segments
    • Or press Ctrl+Shift+T for Translate all not-started & pre-translated

The Batch Translate dialog

The dialog asks how the segments should be translated. Tick one of:

  • 📖 TM (Translation Memory) – pre-translate from the TMs switched on for the project (exact matches and fuzzy matches from 75%). Tick ⚡ Exact matches only to use 100% matches only, which is fastest.
  • 🤖 LLM (AI) – translate with AI (the default). The provider and model are the ones chosen in Settings → AI Settings; the dialog shows which one under 📊 Current LLM.
  • 🌐 MT (Machine Translation) – use the MT providers you have set up in Settings → MT Settings.

Options

  • 🔄 Retry until all segments are translated (recommended) – see Retry Feature below.
  • ✔ Auto-confirm 100% TM matches – with TM selected, exact matches are confirmed straight away instead of getting the TM 100% status.
  • 🔧 Use FuzzyFixer (adapt fuzzy TM matches with AI) – see FuzzyFixer. Segments are then sent one at a time.

Click Start Translation to begin.

The cost estimate

With 🤖 LLM (AI) ticked, the dialog shows what the run is likely to cost before you start, for example:

💰 Estimated cost: ~$0.42 · ~120,000 tokens in / ~35,000 out in 12 call(s)
  • The estimate uses your actual prompt and glossary, your Batch size (Settings → AI Settings) and the same price list as the usage log. When the prompt is long enough for the provider to cache it, batches 2 onwards are priced at the cheaper cached rate, as they will be billed.
  • It updates when you tick 🔧 Use FuzzyFixer, which sends one call per segment and so costs more.
  • It is hidden when 📖 TM or 🌐 MT is ticked.
  • A local (Ollama) model shows as free. A model that isn’t in the price list shows “unknown” rather than pretending to be free.
  • The amount is shown in dollars or euros, whichever you chose in Token Usage & Costs.

It is an estimate (token counts are worked out at about 4 characters per token); the actual cost of every call is logged in Tools → 💰 Token Usage & Costs.

Progress Tracking

During translation:

  • Progress bar shows completion
  • Per-segment status updates
  • Can cancel anytime

Retry Feature

🔄 Retry until all segments are translated is on by default. It will:

  • Automatically detect empty translations
  • Retry failed segments (up to 5 passes)
  • Ensure all segments get translated

Tips

Optimal Batch Size

  • 50-100 segments per batch works well
  • Very large batches may timeout
  • Split by page if needed

Quality vs Speed

  • Claude Sonnet 4.6: Good all-round balance of speed and quality
  • GPT-5.5 / Claude Opus 4.8: Highest quality, slower and more expensive
  • GPT-5.4 Mini / Gemini 3.1 Flash-Lite: Fastest, lower cost

Post-Edit Strategy

After batch translation:

  1. Review each segment
  2. Fix any obvious errors
  3. Confirm with Ctrl+Enter

See Also

🤖 These docs are AI-readable: llms.txt · full · Workbench only